Ordinal Scale Research Articles

We address the problem of predicting a target ordinal variable based on observable features consisting of functional profiles. This problem is crucial, especially in decision-making driven by sensor systems, when the goal is to assess an ordinal variable such as the degree of deterioration, quality level, or risk stage of a process, starting from functional data observed via sensors. We purposely introduce a novel approach called functional-ordinal Canonical Correlation Analysis (foCCA), which is based on a functional data analysis approach. FoCCA allows for dimensionality reduction of observable features while maximizing their ability to differentiate between consecutive levels of an ordinal target variable. Notably, foCCA does not require numerical optimization and is solved in closed form, ensuring computational efficiency and guaranteeing global optimality. FoCCA fully incorporates the ordinal nature of the target variable, embedding it in the Guttman space: this enables the model to capture and represent the relative dissimilarities between consecutive levels of the ordinal target, while also explaining these differences through the functional features. Extensive simulations, and a case study involving the prediction of antigen concentration levels from optical biosensor signals demonstrate the superior performance of foCCA.

Introduction: Patients with return of circulation after an out of hospital cardiac arrest (OHCA) remain at high risk for in-hospital morbidity and mortality from multiorgan dysfunction or anoxic brain injury. Prior studies have reported factors associated with outcomes in OHCA. However, these studies do not address early prognostication factors in the first 24 hours of admission. Outcome predictors available in the initial 24 hours may be helpful in deciding candidacy for advanced therapies such as ECMO and in informing families about potential outcomes. Research question: Evaluate parameters available in the initial 24 hours after hospitalization and their relation to outcomes of children admitted to the Pediatric Intensive Care Unit (PICU) after OHCA. Methods: Single center retrospective study of PICU admissions after OHCA from 2018 to 2024. Data sources were the local Virtual Pediatric Systems registry and the electronic medical records. Pertinent categories included demographics, etiology of OHCA, CT Head findings on admission, neurological pupillary index (NPi) at 0, 6, 12 and 24-hours from admission, serum lactate, base deficit, illness severity score, CPR characteristics, and Pediatric Cerebral Performance Category (PCPC) score. Poor outcome was defined as change in PCPC ≥ 2. Factors associated with outcomes are presented using descriptive statistics, and the sensitivity, specificity, positive and negative predictive values of their association with outcomes were evaluated. Nominal and ordinal variables were compared using Fisher’s exact test and Mann Whitney U test respectively. Results: In this cohort of 218 patients, median age was 2.2 years (0.6, 9.7), 57.3% were males and mortality was 39%. Of the 123 head CTs obtained upon admission, 50 showed cerebral edema (CE), and 88% had poor outcomes (specificity 88.2%, positive predictive value (PPV) 88%). Of the poor outcomes, 80% died and 8% survived. Similarly, a NPi of 0 at all time intervals (0, 6, 12 and 24 hours) was associated with poor outcomes (specificity 96-97% and PPV 96-97%). A cardiac etiology of arrest was associated with survival (79%, p=0.04). Non-survivors had higher illness severity, more epinephrine doses, worse lactate and base deficit values, and higher vasoactive scores. Conclusion: Clinical characteristics in the first 24 hours after OHCA are associated with outcomes. CE on initial head CT and NPi=0 during the initial 24 hours had high specificity and PPV for poor outcomes.

Introduction: Simulation based education (SBE) is integral to the Internal Medical trainee (IMT) curriculum (1). As a centre, we have developed a run-through series of courses delivered annually to IMT doctors across the region in years 1-7 of their training- IMTSim. The learning objectives for these courses are curriculum mapped and incorporate spiral learning to build on key topics. Data collected through pre- and post- course questionnaires suggest that trainees find our courses enjoyable and beneficial to their professional development. As data on the longer term benefits of SBE is limited, we felt it important to evaluate the ongoing impact of our courses via a ‘one year on’ impact survey. Methods: A ‘one year on’ survey was developed for each of our individual IMT courses, allowing the questions to be specific to learning outcomes at different levels of training, and distributed to all doctors who attended our courses between August 2022 and 2024. Questions focused on the application of candidates’ learning during IMTSim to their every-day practice, and their thoughts on SBE as a whole. Qualitative data underwent thematic analysis by two individuals. Quantitative Ordinal Likert scale data was analysed using non-parametric statistical tests. Results: ‘On the day’ surveys showed a significant difference in pre- and post-course ratings of knowledge of human factors, non-technical skills and the role of debriefing, as well as confidence ratings across a range of skills appropriate to specific learning outcomes at different levels of training. Around 200 doctors attend our IMTSim courses each year, and a total of 60 respondents contributed to our follow-up impact survey, with significant numbers reporting use of the skills/ themes explored during our courses in their everyday practice. When asked about SBE as a whole, significant numbers reported that they felt it was more impactful (87%) and more focused on the individual learner (78%) compared to more traditional teaching modalities. Discussion: Our data demonstrates immersive simulation has longer term impact on IMT doctors. Learning continued to be retained at one year post-course, with individuals going on to use and implement skills learned within their routine clinical practice. Ethics Statement: As the submitting author, I can confirm that all relevant ethical standards of research and dissemination have been met. Additionally, I can confirm that the necessary ethical approval has been obtained, where applicable.

Introduction: Circadian misalignment is an emerging risk factor for poor cardiovascular health, and chronotype may reflect underlying circadian processes. While previous studies have primarily reported cross-sectional associations between chronotype and individual cardiovascular risk factors, Mendelian randomization (MR) may provide further insights into the role of chronotype in overall cardiovascular health, as measured by the American Heart Association’s Life’s Essential 8 (LE8). Hypothesis: We hypothesize that individuals with a morning chronotype have better overall cardiovascular health measured by LE8 scores. Methods: Cross-sectional and one-sample MR analyses included 272,571 UK Biobank (UKB) participants of White European ancestry. Chronotype was self-reported using a five-level ordinal scale modeled linearly from -2 (evening) to 2 (morning). An unweighted genetic score from 341 chronotype-associated SNPs from a UKB GWAS served as the MR instrument (R 2 = 2.2%). Two-stage least-squares regression estimated difference in LE8 per one-unit increment in chronotype, adjusting for age, sex, assessment center, genotyping batch, and 40 principal components (PCs). To mitigate winner’s curse bias (inflated GWAS estimates), we replicated the analysis in 11,217 White women from the Nurses’ Health Study II (NHSII), using a weighted genetic score (R 2 = 1.4%) and a reduced LE8 score based on six self-reported components (diet, physical activity, nicotine exposure, sleep, body weight, and blood pressure). Results: In UKB, each one-unit increment toward more morningness was associated with a 0.75-point higher LE8 score for self-reported chronotype (95% CI: 0.72, 0.78; P < 2.2×10 -16 ) and a 0.76-point higher LE8 score for genetically predicted chronotype (95% CI: 0.54, 0.98; P = 1.8×10 -11 ). Observational and MR results were slightly stronger in women. In NHSII, self-reported morning chronotype was associated with a higher LE8 score (β = 1.47, 95% CI: 1.28, 1.67; P = 3.32×10 -52 ). However, the MR estimate, while positive, did not reach statistical significance (β = 1.27, 95% CI: -0.44, 2.98; P = 0.14). When examining individual components, morning chronotype was consistently associated only with diet score across both observational and MR analyses and in both cohorts. Conclusions: Our findings suggest a causal relationship between morning chronotype and better cardiovascular health, particularly diet quality, although replication in other populations remains necessary.

Table tennis (TT) is recognized for its accessibility, adaptability, and health benefits, making it suitable for physical education (PE). This study aimed to investigate the prevalence of TT implementation in primary school PE classes and explored associations with teacher characteristics and perceptions regarding injury risk, safety, inclusivity for students with disabilities or special educational needs, student engagement, and the educational value in PE curricula. A cross-sectional survey was conducted among 393 Spanish primary school PE teachers who completed the validated Racket Sports Attitude Scale (RSAS). Associations between teacher characteristics and TT use were tested using Pearson’s Chi-square, with effect sizes reported as Cramer’s V, Phi, and Somers’ D for ordinal variables. Additional analyses examined links between TT adoption and perceptions of injury risk, inclusivity, safety for pupils, ability to enhance engagement, and its educational value in PE. Only 11.7% of teachers reported using TT in PE classes. Implementation rates were not significantly different by sex but were associated with age (χ2 = 27.2, p < 0.001, Somers’ D = 0.071) and teaching experience (χ2 = 30.0, p < 0.001, Somers’ D = 0.099). TT use showed strong associations with perceptions of lower injury risk (Cramer’s V = 0.707), suitability for students with disabilities (0.712), special educational needs (0.715), safety (0.707), engagement (0.712), and educational value (0.716) (p < 0.001). Despite positive perceptions, TT is underutilized in PE curricula.

Alteplase remains the guideline-directed fibrinolytic of choice for acute ischemic stroke (AIS), however, tenecteplase has grown in popularity as new data have emerged. The purpose of this study is to compare door-to-needle (DTN) times for three fibrinolytic strategies: alteplase as drug of choice, tenecteplase as drug of choice, and choice of alteplase/tenecteplase based on institution use criteria. This was a single-center retrospective cohort study including patients 18 years and older who presented to the emergency department and received alteplase or tenecteplase for AIS from November 1, 2018 to December 31, 2023. The primary outcome was DTN time calculated from emergency department arrival until fibrinolytic administration. Continuous data were assessed using a one-way ANOVA with Tukey HSD. Nominal data were assessed using a Chi-square test, and ordinal data were assessed using a Kruskal-Wallis test. A total of 240 patients were evaluated for the study, with 80 patients in each of the three groups. There was no difference in DTN times between the alteplase, tenecteplase/alteplase, and tenecteplase groups (39, 47, 48 minutes, P = .25). The time from start of CT scan to fibrinolytic administration was significantly shorter between the alteplase and tenecteplase/alteplase groups (27, 38 minutes, P = .01). This project highlighted the potential effects of having more than one fibrinolytic agent on formulary for acute ischemic stroke with different criteria for use. DTN time was not prolonged with two fibrinolytics on formulary.

This meta-analytical review critically examines the Ordinal Logistic Regression (OLR) model, focusing on its applicability, underlying assumptions and practical constraints in empirical research. Drawing on Forty-five (45) peer-reviewed studies published between 2015 and 2025, the analysis synthesizes methodological insights from applications across diverse fields, including social sciences, health research, education and engineering. The review outlines the theoretical foundation of OLR, highlighting its suitability for modeling ordinal response variables where the proportional odds assumption holds. Key assumptions such as proportionality of odds, absence of multicollinearity, and correct model specification are systematically discussed, alongside diagnostic procedures for verification. Practical constraints identified include small sample bias, violation of proportional odds, challenges in interpreting interaction effects and the complexity of handling missing or imbalanced data. The findings underscore that while OLR offers robust analytical power for ordinal outcomes, its validity is contingent upon careful assumption testing, adequate sample size, and context-appropriate interpretation. This synthesis contributes to improved methodological rigor by offering a consolidated reference for researchers, emphasizing both the strengths and limitations of the OLR model in applied statistical modeling.

Reading systemic disease from the feline fundus: A multicenter comparison of multimodal large language models and veterinary clinicians.

Skeletal metastasis is not only associated with pain but also constitutes a prevalent cause of mortality in patients with prostate cancer (PC). Accurate evaluation of skeletal disease status is essential in the management of PC. This study examines the concordance between 99mTc-prostate-specific membrane antigen (PSMA) and 99mTc-Methylene Diphosphonate (MDP) scans for detecting skeletal metastases in PC patients. This prospective study evaluated 18 participants with histopathologically confirmed PC from July 2022 to July 2024. All patients underwent 99mTc-PSMA and 99mTc-MDP single-photon emission computed tomography/computed tomography (SPECT/CT) within an average interval of 9.5 days. Lesions were categorized on an ordinal scale. The Wilcoxon signed-rank test was employed for statistical comparison. The median prostate-specific antigen (PSA) at the time of the study was 85.9ng/ml. The disease was predominantly advanced, with 14 patients (74%) exhibiting stage 4b. The Gleason scores were 8-10 (n = 16) and 7 (n = 2). Total 244 skeletal lesions were detected. 99mTc-MDP identified 233 lesions (95.5%), whereas 99mTc-PSMA identified 176 lesions (72.1%). The disparity in lesion count across scans was not statistically significant (Z = 1.6, P = 0.11). Area under the curve for receiver operating characteristic analysis of 99mTc-PSMA for skeletal lesion detection was 0.878, signifying high diagnostic precision. The 99mTc-PSMA scan exhibited comparable efficacy to 99mTc-MDP bone scintigraphy in identifying skeletal metastases in PC. The capability to identify both skeletal and supplementary soft tissue metastases, together with its potential theragnostic applications, establishes 99mTc-PSMA scan as a viable comprehensive imaging solution, especially in resource-constrained settings with limited PSMA PET/CT access and poor patient affordability.

Bifocal Surgery for Malignant Spinal Lesions is Associated with Shorter Postoperative Survival and Higher Postoperative Complication Rates Compared to Monofocal Surgery.

Abstract Indirect calorimetry is a cornerstone technique for metabolic phenotyping of animal models in pre-clinical research, with well-established experimental protocols and platforms. However, a flexible, extensible, and user-friendly software suite that enables standardized integration of data and metadata from diverse metabolic phenotyping platforms—followed by unified statistical analysis and visualization—remains absent. We present Shiny-Calorie, an open-source interactive application for transparent data and metadata integration, comprehensive statistical data analysis, and visualization of indirect calorimetry datasets. Shiny-Calorie supports the majority of standard data formats across commercial metabolic phenotyping platforms, such as TSE and Sable Systems, COSMED platform and CLAMS/Columbus instruments, including export functionality of processed data into standardized data formats. Built using GNU R and a reactive interface, Shiny-Calorie enables intuitive exploration of complex, multi-modal longitudinal datasets comprising categorical, continuous, ordinal, and count variables. The platform incorporates state-of-the-art statistical methods for robust hypothesis testing, thereby facilitating biologically meaningful interpretation of energy metabolism phenotypes, including resting metabolic rate and energy expenditure. Overall, Shiny-Calorie streamlines routine analysis workflows and enhances reproducibility and transparency in metabolic phenotyping studies.

Background Simulation-based learning is an effective tool for educating teachers and school staff about anaphylaxis. This pilot exploratory study aimed to assess knowledge and attitudes toward food allergy and anaphylaxis policies and protocols in schools for the management of severe allergic reactions. Materials and methods A simulation-based education program was conducted using a standardized case scenario in elementary schools in Rabigh, targeting teachers and staff from randomly selected schools. Knowledge and attitudes were assessed before and after the simulation using structured questionnaires. Paired pre–post responses ( n = 97) were analyzed. McNemar’s test was applied for categorical outcomes, and the Wilcoxon signed-rank test was used for ordinal attitude scores. Data were summarized as frequencies and percentages, with a significance level set at p < 0.05. Results A total of 101 teachers and staff from six elementary schools participated, with 97 completing both pre- and post-simulation assessments. Recognition of key anaphylaxis symptoms improved significantly post-intervention, particularly for swelling of the lips, tongue, face, and eyes (55.4% vs. 79.4%) and shortness of breath (28.7% vs. 78.4%) (both p < 0.001, McNemar’s test). Attitudes toward anaphylaxis management protocols also improved, with a significant reduction in median attitude scores post-simulation ( p = 0.001, Wilcoxon signed-rank test). Readiness to use an epinephrine auto-injector increased markedly from 18.8% before the intervention to 84.5% after the intervention ( p < 0.001). Conclusion This pilot exploratory study demonstrates that simulation-based education can significantly improve teachers’ knowledge, attitudes, and readiness to manage food allergy and anaphylaxis. While the findings are promising, they should be interpreted with caution due to the modest sample size and single-city setting. Larger, multi-center studies.

Purpose This study aims to evaluate the development of complex thinking among Filipino university students and to validate a hierarchical measurement model for this competency, which is essential for navigating the country's digital, economic, and sustainability challenges. Design/methodology/approach The eComplexity instrument was applied to a sample of 205 students from diverse disciplines. Structural equation modeling (SEM) with Diagonally Weighted Least Squares (DWLS) estimation was used, suitable for ordinal data. The model included four subdimensions: systemic, scientific, critical, and innovative thinking. Findings Results showed excellent model fit (CFI = 0.992, TLI = 0.991, SRMR = 0.066), high reliability (CR = 0.837–0.912), and strong explained variance. Scientific thinking emerged as the most solid subdimension, while critical and innovative thinking revealed lower item consistency, indicating areas for pedagogical improvement. Research limitations/implications The study's scope is limited by its sample size and institutional context, restricting the generalizability of results. Future research should include longitudinal studies and expanded samples to explore disciplinary and institutional variations in complex thinking development. Practical implications The findings underscore the need to reinforce critical and innovative thinking through pedagogical strategies that promote metacognition and real-world problem-solving. This can guide curriculum reforms in Philippine higher education aimed at building transversal competencies. Originality/value This study offers empirical validation of a hierarchical model for complex thinking and provides actionable insights for integrating this competency into higher education. It contributes to educational transformation by aligning curriculum development with national and global demands.

This study examines the effect of multicollinearity on ordinal regression through a two-stage Monte Carlo simulation. A synthetic population of 2,000,000 observations was generated with predictors drawn from a normal distribution, and responses simulated using an ordinal probit model. A Monte Carlo procedure was employed with 10 repetitions, each consisting of 100 random samples of 1,000 observations. Parameter estimation employed Maximum Likelihood Estimation (MLE) for univariate models and Pairwise Likelihood (PL) for multivariate models, with performance assessed using mean squared error (MSE), bias, and computation time. Results show that multicollinearity had negligible impact on estimator bias and MSE, confirming the robustness of both MLE and PL to correlated predictors. However, severe multicollinearity substantially increased computation time, indicating a trade-off between estimator stability and efficiency. These findings highlight PL as a viable approach for analyzing complex ordinal data, particularly in applications such as socio-economic surveys and health metrics where predictor correlation is unavoidable.

Work-related stress has emerged as a critical issue in public health and labour policy, particularly in light of evolving work environments and the effects of the COVID-19 pandemic. This study examines the perception of work stress across European Union member states by utilising CUB models, a class of mixture models well-suited for analysing ordinal data. The data for this research stems from the Flash Eurobarometer - OSH Pulse survey, conducted by the European Commission Agency for Occupational Safety and Health (EU-OSHA) in Spring 2022. Our findings indicate that the pandemic crisis has affected workers in the EU differently, with a greater impact on self-employed individuals, women, and more educated workers, who reported higher levels of work-stress. To support this analysis, the graphical features of CUB models effectively illustrate diverse response patterns across different countries and respondent profiles. Additionally, the Gretl CUB package has been updated to allow the use of sample weights in the estimation algorithm.

The aim of this study was to assess the effectiveness of adjunct percutaneous transluminal angioplasty and stenting (PTAS) in endovascular thrombectomy (EVT) for large vessel occlusion (LVO) in acute ischemic stroke with underlying intracranial atherosclerotic disease (ICAD). Data from 2019 to 2023 were collected from a multicenter national EVT registry in Taiwan. Patients with ICAD were classified into PTAS and non-PTAS groups according to the adjunct therapy administered. The efficacy and safety outcomes of interest included successful reperfusion, the 90-day modified Rankin Scale (mRS) score, mortality, symptomatic intracerebral hemorrhage (ICH), and EVT complication. Multivariable logistic regression and inverse probability of treatment weighting-adjusted analyses were performed to compare outcomes. A total of 585 patients with ICAD were included in the study, with 94 in the PTAS group and 491 in the non-PTAS group. PTAS was associated with a higher successful reperfusion rate as the primary efficacy outcome (89.4% vs 75.0%; adjusted OR 2.80 [95% CI 1.37-5.75], p < 0.01). This effect was preserved even in the pure angioplasty group (unadjusted OR 2.44 [95% CI 1.08-5.51], p = 0.03; adjusted OR 2.61 [95% CI 1.13-6.08], p = 0.03). Among secondary outcomes, the PTAS group had better ambulatory functional outcomes (mRS score 0-3) (53.7% vs 40.4%; weighted OR 1.53 [95% CI 1.20-1.95], p < 0.01). There was no significant difference in independent functional outcomes (mRS score 0-2) or ordinal 90-day mRS scores. For safety outcomes, symptomatic ICH and EVT-related complications were comparable between groups. No interaction effect was found for the onset-to-puncture time and PTAS. Subgroup analyses revealed that PTAS offered more benefits for posterior circulation strokes. This study highlights PTAS as complementary therapy in EVT for ICAD-related LVO, with consistent effects across various time metrics (onset-to-puncture time) and no significant increase in adverse events.

Influence of head and neck radiotherapy doses on apical periodontitis progression: An animal-based study.

The study evaluated neonatal cerebro-spinal fluid (CSF) leucocyte esterase (LE) estimation by reagent strips to diagnose CSF leucocytosis insituations where the CSF leucocyte count is delayed or unavailable. This prospective diagnostic study had a cohort-selection design. CSF samples of neonates with clinical sepsis were analysed for LE by reagent strips, cytochemistry, gram stain and culture. CSF LE was measured on an ordinal scale from nil to large. Correlations of CSF LE with CSF leucocyte and neutrophil count were determined. CSF LE was compared versus CSF leucocytosis (defined a priori ≥ 25 leucocytes/μL) and versus CSF neutrophilia (defined a priori ≥ 15 neutrophils/μL) as the reference standard. A total of 341 neonates were enrolled. CSF LE values correlated with leucocyte count (rho = 0.49, p < 0.001) and neutrophil count (rho = 0.44, p < 0.001). LE had a sensitivity (95% CI) of 63.5% (51.2, 74.1), specificity 90.7% (88.4, 92.6), LR+ (likelihood ratio) 6.79 (4.43, 9.98) and LR- 0.40 (0.28, 0.55) to diagnose CSF leucocytosis. When expressed as an ordinal variable, LE had an area under the receiver operator characteristic curve of 0.77 (95% CI: 0.69, 0.85; p < 0.001) to detect leucocytosis. With respect to all parameters, the diagnostic performance of CSF LE to detect neutrophilia was inferior to the detection of leucocytosis. CSF leucocyte esterase estimation by reagent strips has a moderately high specificity and a good LR+ but, overall, does not have sufficiently robust diagnostic accuracy to substitute CSF leucocyte counts.

Patient-centered care for ulcerative colitis (UC) involves reducing the burden associated with colonoscopy, particularly bowel preparation. Although low-volume preparations have become common in the general population, data on the use of 1-L polyethylene glycol with ascorbic acid (PEG+Asc) in patients with UC are limited. We compared the efficacy, safety, and acceptability of 1-L and 2-L PEG+Asc in patients with quiescent UC. This was a multicenter, randomized, single-blind, non-inferiority study. Adult outpatients with UC who had stable disease activity were randomly allocated to 1-L or 2-L of PEG+Asc for colonoscopy. Degree of bowel cleansing was assessed using the Boston Bowel Preparation Scale and rated as successful cleansing if the score was ≥6 with all segment scores ≥2. Patient acceptance (ease of administration and willingness to repeat) and tolerability (newly developed symptoms, such as nausea, bloating, and abdominal pain) were assessed using a four-point ordinal scale. Disease activity (partial or full Mayo score) and laboratory data before and after colonoscopy were assessed for safety concerns. Of the 196 randomized patients, 74 in the 1-L group and 70 in the 2-L group, respectively, completed the study. Successful cleansing was achieved in 98.6% and 97.1% of patients, respectively (absolute difference -1.4%, one-sided 97.5% confidence interval -5.9%), meeting the non-inferiority margin in the per-protocol analysis. A conservative intention-to-treat analysis did not meet the non-inferiority threshold. Overall tolerability and acceptability were similar, although nausea was reported more frequently in the 1-L group. No significant changes in disease activity were observed, and minor electrolyte shifts occurred more often in the 1-L group but were clinically insignificant. One-liter PEG+Asc is effective and safe for bowel preparation in patients with quiescent UC and offers a viable low-volume alternative to the 2-L regimen. Careful patient selection and monitoring may be advisable in elderly or comorbid patients.

Despite limited research, cancer patients are opting for compounds found in cannabis, like tetrahydrocannabinol (THC) and cannabidiol (CBD), to improve their sleep. The purpose of this study was to examine the therapeutic value of cannabis for sleep. Patient-reported symptom responses were obtained from 1962 cancer patients enrolled in the Minnesota Medical Cannabis Program (MMCP) from 2015 to 2023. Multivariable logistic and linear regression models were used to evaluate the associations between changes in reported sleep disturbance scores and the dose of THC, the dose of CBD, and the cannabinoid ratio (THC:CBD). Logistic and linear regression models were adjusted for sex, age, race, ethnicity, body mass index, and MMCP enrollment fee category. Linear regression models were additionally adjusted for baseline sleep disturbance score. Compared to the highest quintile category of CBD dose, lower dose quintiles were 29-35% less likely to be associated with at least a 30% improvement in sleep disturbance scores. Sleep disturbance scores improved by 1.87 points on a 0-10 ordinal scale for cancer patients with CBD doses in the top quintile, and approximately 1.5 points for doses in lower quintiles. THC and THC:CBD doses were not consistently related to changes in sleep disturbance scores. Higher CBD doses may be associated with clinically meaningful improvements in sleep in cancer patients enrolled in a medical cannabis program.