Category: Statistics

Cognition. 2024 Nov 14;254:106002. doi: 10.1016/j.cognition.2024.106002. Online ahead of print.

ABSTRACT

While the visual world is rich and complex, importantly, it nevertheless contains many statistical regularities. For example, environmental feature distributions tend to remain relatively stable from one moment to the next. Recent findings have shown how observers can learn surprising details of environmental color distributions, even when the colors belong to actively ignored stimuli such as distractors in visual search. Our aim was to determine whether such learning influences orienting in the visual environment, measured with saccadic eye movements. In two visual search experiments, observers had to find an odd-one-out target. Firstly, we tested cases where observers selected targets by fixating them. Secondly, we measured saccadic eye movements when observers made judgments on the target and responded manually. Trials were structured in blocks, containing learning trials where distractors came from the same color distribution (uniform or Gaussian) while on subsequent test trials, the target was at different distances from the mean of the learning distractor distribution. For both manual and saccadic measures, performance improved throughout the learning trials and was better when the distractor colors came from a Gaussian distribution. Moreover, saccade latencies during test trials depended on the distance between the color of the current target and the distractors on learning trials, replicating results obtained with manual responses. Latencies were slowed when the target color was within the learning distractor color distribution and also revealed that observers learned the difference between uniform and Gaussian distributions. The importance of several variables in predicting saccadic and manual reaction times was studied using random forests, revealing similar rankings for both modalities, although previous distractor color had a higher impact on free eye movements. Overall, our results demonstrate learning of detailed characteristics of environmental color distributions that affects early attentional selection rather than later decisional processes.

PMID:39546817 | DOI:10.1016/j.cognition.2024.106002

JMIR Form Res. 2024 Nov 15;8:e56006. doi: 10.2196/56006.

ABSTRACT

BACKGROUND: Studying illicit drug circulation and its effects on population health is complicated due to the criminalization of trade and consumption. Illicit drug markets have evolved with IT, moving digital to the “darknet.” Previous research has analyzed darknet market listings and customer reviews. Research tools include public health surveys and medical reports but lack neutral data on drugs’ spread and impact. This study fills this gap with an analysis of the volume of drugs traded on the darknet market.

OBJECTIVE: We aimed to use the dark web data and officially published indicators to identify the most vulnerable regions of Russia and the correlations between the pairs of variables to measure how illicit drug trade can affect population well-being.

METHODS: We web-parsed the Hydra darknet drug marketplace using Python code. The dataset encompassed 3045 individual sellers marketing 6721 unique products via 58,563 distinct postings, each representing specific quantities sold in different Russian regions during 2019. In the second stage, we collected 31 variables from official sources to compare officially collected data with darknet data about amounts and types of selling drugs in every 85 regions of Russia. The health-related data were obtained from official published sources-statistical yearbooks. Maps, diagrams, correlation matrixes, and applied observational statistical methods were used.

RESULTS: In 2019, a minimum of 124 kilograms of drugs circulated daily in small batches on the Russian darknet. Cannabis dominated the market, being 10 times more prevalent than opiates, and cannabis products’ higher availability in the region is correlated with a lower incidence of opiate overdoses. The “grams of opiates in the region” variable is significantly correlated with drug overdose deaths (r=.41; P=.003), HIV-positive cases due to drug use (r=.51; P=.002), and drug court convictions in Russia (r=.39; P=.004). The study identified significant correlations between opiate sales on the darknet and higher rates of HIV among injection drug users (r=.47; P=.003). Conversely, regions with higher cannabis sales exhibited significant negative correlations with indicators of harmful drug use (r=-.52; P=.002) and its prevalence (r=-.49; P=.001). These findings suggest regional variations in drug sales on the darknet may be associated with differing public health outcomes. These indicators accurately reflect regional drug issues, though some official statistics may be incomplete or biased.

CONCLUSIONS: Our findings point to varying levels of risk associated with different types of drugs sold on the darknet, but further research is needed to explore these relationships in greater depth. The study’s findings highlight the importance of considering regional variations in darknet drug sales when developing public health strategies. The significant correlations between drug sales data and public health indicators suggest that region-specific interventions could be more effective in addressing the diverse challenges posed by illicit drug use.

PMID:39546792 | DOI:10.2196/56006

J Neurosurg. 2024 Nov 15:1-10. doi: 10.3171/2024.7.JNS24373. Online ahead of print.

ABSTRACT

OBJECTIVE: There is persistent debate in the literature surrounding the true predictors of biochemical remission after resection of somatotroph adenoma. A multimodal analysis of a large number of patients is needed to better understand which patients may be at higher or lower risk for remission failure after surgery.

METHODS: A retrospective review was performed on patients undergoing somatotroph adenoma resection. Biochemical remission was defined as age- and sex-adjusted normalization of serum insulin growth factor-1 (IGF-1) levels at least 6 months after surgery. Patient case characteristics and clinicopathologic variables were tested for statistical associations with remission and were included in a random forest machine learning model to assess for their importance in determining remission status. Preoperative variables found to be significant remission predictors on statistical testing and important in the random forest model were subsequently assessed via receiver operating characteristic (ROC) analysis to determine numeric thresholds that optimally predicted preoperative likelihood of remission success or failure.

RESULTS: Eighty patients were identified with somatotroph adenoma who underwent transsphenoidal resection, with 60 patients (75%) achieving biochemical remission. Statistical testing found that patients with failed remission were more likely to have larger tumors (1.9 vs 1.6 cm by the largest axis, p = 0.014; and 3.61 vs 2.66 cm3 by 3D volume, p = 0.013) that invaded the cavernous sinus more frequently (70% vs 22% of patients, p < 0.001) and have higher preoperative IGF-1 level (860 vs 660 ng/ml, p = 0.044). An optimized random forest machine learning model with 10,000 iterations found that tumor size, preoperative growth hormone and IGF-1 levels, and cavernous sinus invasion were important preoperative predictors of remission status. ROC analysis revealed that 96% of patients with preoperative 3D tumor volume less than 1.51 cm3 (area under the curve [AUC] 0.691, p = 0.003) and 100% with nonadjusted preoperative IGF-1 level less than 718.5 ng/ml (AUC 0.736, p = 0.002) achieved remission.

CONCLUSIONS: Important preoperative predictors of postoperative remission for somatotroph adenoma resection include serum IGF-1 level, cavernous sinus invasion, and tumor size. Ninety-five percent of patients who achieved postoperative remission had preoperative 3D tumor volume less than 1.51 cm3.

PMID:39546789 | DOI:10.3171/2024.7.JNS24373

J Neurosurg. 2024 Nov 15:1-8. doi: 10.3171/2024.6.JNS232599. Online ahead of print.

ABSTRACT

OBJECTIVE: The authors’ goal was to perform a retrospective audit of all emergency cranial neurosurgery performed at the Royal Darwin Hospital in the first 5 years of the unit and to compile their data in a similar fashion to an earlier study titled “Emergency Neurosurgery in Darwin: Still the Generalist Surgeons’ Responsibility,” which was published in 2015.

METHODS: All emergency cranial neurosurgery performed by a neurosurgeon between 2017 and 2021 was identified. Data were extracted from the National Critical Care and Trauma Response Centre database. Statistical analyses were descriptive logistic regression performed using Stata version 15.1 software to examine factors associated with death.

RESULTS: A total of 320 patients (42% Indigenous) underwent 427 emergency neurosurgeries. There were 35 emergency neurosurgeries in 2017 and 82 in 2021. The most common procedure performed was insertion of an external ventricular drain, followed by craniotomy and removal of intracranial hematoma. Mortality was 7.5% overall and 8.4% among patients with trauma. Only age proved to be a statistically significant independent risk factor for death (t = -2.95, p < 0.0041; OR 1.06, p = 0.02). Location, sex, injury severity, and presenting Glasgow Coma Scale score were not associated with death. Indigenous and non-Indigenous patients had similar outcomes.

CONCLUSIONS: The data illustrate the importance of developing small but sustainable neurosurgical units in rural and remote areas. A dedicated neurosurgical unit at the Royal Darwin Hospital has led to an increase in the amount and variety of emergency neurosurgery performed in Darwin. Interstate transfers have reduced. This has tangible lifesaving and economic advantages.

PMID:39546787 | DOI:10.3171/2024.6.JNS232599

JMIR Res Protoc. 2024 Nov 15;13:e58140. doi: 10.2196/58140.

ABSTRACT

BACKGROUND: Antimicrobial resistance (AMR) is a major global public health concern, particularly in low- and middle-income countries where resources and infrastructure for an adequate response are limited. The World Health Organization (WHO) Global Antimicrobial Resistance Surveillance System (GLASS) was introduced in 2016 to address these challenges, outlining recommendations for priority pathogen-antibiotic combinations. Despite this initiative, implementation in Africa remains understudied. This scoping review aims to assess the current state of implementing WHO GLASS recommendations on antimicrobial sensitivity testing (AST) in Africa.

OBJECTIVE: The primary objective of this study is to determine the current state of implementing the WHO GLASS recommendations on AST for priority pathogen-antimicrobial combinations. The review will further document if the reporting of AST results is according to “susceptible,” “intermediate,” and “resistant” recommendations according to GLASS.

METHODS: Following the methodological framework by Arksey and O’Malley, studies published between January 2016 and November 2023 will be included. Search strategies will target electronic databases, including MEDLINE, Scopus, CINAHL, and Embase. Eligible studies will document isolates tested for antimicrobial sensitivity, focusing on WHO-priority specimens and pathogens. Data extraction will focus on key study characteristics, study context, population, and adherence to WHO GLASS recommendations on AST. Descriptive statistics involving summarizing the quantitative data extracted through measures of central tendency and variation will be used. Covidence and Microsoft Excel software will be used. This study will systematically identify, collate, and analyze relevant studies and data sources based on clear inclusion criteria to provide a clear picture of the progress achieved in the implementation of the WHO GLASS recommendations. Areas for further improvement will be documented to inform future efforts to strengthen GLASS implementation for enhanced AMR surveillance in Africa.

RESULTS: The study results are expected in August 2024.

CONCLUSIONS: To our knowledge, this scoping review will be the first to comprehensively examine the implementation of WHO GLASS recommendations in Africa, shedding light on the challenges and successes of AMR surveillance in the region. Addressing these issues aims to contribute to global efforts to combat AMR.

INTERNATIONAL REGISTERED REPORT IDENTIFIER (IRRID): PRR1-10.2196/58140.

PMID:39546786 | DOI:10.2196/58140

J Neurosurg Spine. 2024 Nov 15:1-9. doi: 10.3171/2024.7.SPINE24464. Online ahead of print.

ABSTRACT

OBJECTIVE: Surgical correction of adult spinal deformity (ASD) is associated with a high rate of hardware complication that can be challenging to predict. Hardware integrity and alignment after surgery are typically followed with standing radiography, where pedicle screw loosening may be incidentally identified but the clinical significance of which is often unclear. This study aimed to identify the incidence and implications of pedicle screw loosening at the upper instrumented vertebra (UIV) after surgical correction of ASD.

METHODS: A single-institution retrospective analysis was performed on a cohort of 217 patients who underwent long-segment fusion with pelvic fixation for correction of ASD between September 2013 and November 2021. Cases with a minimum 1-year follow-up were included. UIV pedicle screws were graded on radiographs for evidence of loosening with a 0- to 3-point scale: 0, no loosening; 1, lucency within screw threads; 2, lucency around screw threads; and 3, screw dislodgment/backout. Need for hardware revision surgery was assessed as the primary outcome. Patient-reported outcome measures (PROMIS and Oswestry Disability Index scores) were assessed as secondary outcomes among the patients with available scores.

RESULTS: Low-grade UIV screw loosening (grade 1) was identified in 37 patients (17.1%), and high-grade UIV loosening (grade 2 or 3) was identified in 23 patients (10.6%). Low-grade UIV loosening was not associated with eventual need for hardware revision (OR 0.52, 95% CI 0.17-1.61, p = 0.258); however, high-grade loosening was associated with increased odds of hardware revision (OR 5.17, 95% CI 1.74-15.36, p = 0.003), including specifically surgery for correction of proximal junctional kyphosis (OR 5.73, 95% CI 1.27-25.95, p = 0.024). Among patients with PROMIS T-scores, those requiring hardware revision reported worse Pain Interference (65.0 ± 5.1 vs 59.6 ± 7.7, p = 0.001) and Physical Function (33.3 ± 5.6 vs 37.4 ± 7.4; p = 0.011). Patients with high-grade UIV loosening reported higher Oswestry Disability Index scores than those without high-grade loosening (grade 0 or 1), although this failed to reach statistical significance (44.0 ± 8.5 vs 33.7 ± 18.5, p = 0.101).

CONCLUSIONS: Grade 1 UIV pedicle screw loosening may represent a benign incidental finding, whereas high-grade loosening is associated with significantly increased odds of hardware revision surgery. High-grade loosening may also be associated with worse patient-reported disability. The authors’ findings suggest that while low-grade UIV loosening may often be managed expectantly, identification of high-grade UIV pedicle screw loosening on follow-up imaging warrants increased attention and continued surveillance.

PMID:39546785 | DOI:10.3171/2024.7.SPINE24464

J Med Internet Res. 2024 Nov 15;26:e57628. doi: 10.2196/57628.

ABSTRACT

BACKGROUND: Increasing adoption of sensor-based digital health technologies (sDHTs) in recent years has cast light on the many challenges in implementing these tools into clinical trials and patient care at scale across diverse patient populations; however, the methodological approaches taken toward sDHT usability evaluation have varied markedly.

OBJECTIVE: This review aims to explore the current landscape of studies reporting data related to sDHT human factors, human-centered design, and usability, to inform our concurrent work on developing an evaluation framework for sDHT usability.

METHODS: We conducted a scoping review of studies published between 2013 and 2023 and indexed in PubMed, in which data related to sDHT human factors, human-centered design, and usability were reported. Following a systematic screening process, we extracted the study design, participant sample, the sDHT or sDHTs used, the methods of data capture, and the types of usability-related data captured.

RESULTS: Our literature search returned 442 papers, of which 85 papers were found to be eligible and 83 papers were available for data extraction and not under embargo. In total, 164 sDHTs were evaluated; 141 (86%) sDHTs were wearable tools while the remaining 23 (14%) sDHTs were ambient tools. The majority of studies (55/83, 66%) reported summative evaluations of final-design sDHTs. Almost all studies (82/83, 99%) captured data from targeted end users, but only 18 (22%) out of 83 studies captured data from additional users such as care partners or clinicians. User satisfaction and ease of use were evaluated for 83% (136/164) and 91% (150/164) of sDHTs, respectively; however, learnability, efficiency, and memorability were reported for only 11 (7%), 4 (2%), and 2 (1%) out of 164 sDHTs, respectively. A total of 14 (9%) out of 164 sDHTs were evaluated according to the extent to which users were able to understand the clinical data or other information presented to them (understandability) or the actions or tasks they should complete in response (actionability). Notable gaps in reporting included the absence of a sample size rationale (reported for 21/83, 25% of all studies and 17/55, 31% of summative studies) and incomplete sociodemographic descriptive data (complete age, sex/gender, and race/ethnicity reported for 14/83, 17% of studies).

CONCLUSIONS: Based on our findings, we suggest four actionable recommendations for future studies that will help to advance the implementation of sDHTs: (1) consider an in-depth assessment of technology usability beyond user satisfaction and ease of use, (2) expand recruitment to include important user groups such as clinicians and care partners, (3) report the rationale for key study design considerations including the sample size, and (4) provide rich descriptive statistics regarding the study sample to allow a complete understanding of generalizability to other patient populations and contexts of use.

PMID:39546781 | DOI:10.2196/57628

JMIR Form Res. 2024 Nov 15;8:e64476. doi: 10.2196/64476.

ABSTRACT

BACKGROUND: Experimentation is crucial in chemistry education as it links practical experience with theoretical concepts. However, practical chemistry courses typically rely on real laboratory experiments and often face challenges such as limited resources, equipment shortages, and logistical constraints in university settings. To address these challenges, computer-based laboratories have been introduced as a potential solution, offering electronic simulations that replicate real laboratory experiences.

OBJECTIVE: This study examines the effect of virtual laboratories on the academic achievement of undergraduate chemistry students and evaluates their potential as a viable alternative or complement to traditional laboratory-based instruction.

METHODS: A quasi-experimental design was implemented to examine the cause-and-effect relationship between instructional methods and student outcomes. The study involved 60 fourth-year BSc chemistry students from Dilla University, divided into 3 groups: a real laboratory group (n=20), which performed real laboratory experiments; a virtual group (n=20), which used virtual laboratory simulations; and a lecture group (n=20), which received lecture-based instruction. Quantitative data were collected through tests administered before and after the intervention to assess academic performance. The data analysis used descriptive and inferential statistics, such as means and SDs, 1-way ANOVA, the Tukey honestly significant difference test, and independent-sample t tests (2-tailed), with a P value of .05 set for determining statistical significance.

RESULTS: Before the intervention, the results indicated no significant differences in academic achievement among the 3 groups (P=.99). However, after the intervention, notable differences were observed in student performance across the methods. The real laboratory group had the highest mean posttest score (mean 62.6, SD 10.7), followed by the virtual laboratory group (mean 55.5, SD 6.8) and the lecture-only group, which had the lowest mean score (mean 43.7, SD 11.5). ANOVA results confirmed significant differences between the groups (F_2,57=18.429; P<.001). The Tukey post hoc test further revealed that the real laboratory group significantly outperformed the lecture-only group (mean difference 18.88; P<.001), while the virtual laboratory group also performed significantly better than the lecture-only group (mean difference 11.7; P=.001). However, no statistically significant difference was found between the real laboratory and virtual laboratory groups (mean difference 7.12; P=.07). In addition, gender did not significantly influence performance in the virtual laboratory group (P=.21), with no substantial difference in posttest scores between male and female students.

CONCLUSIONS: These findings suggest that computer-based laboratories are a viable and effective alternative when real laboratories are unavailable, enhancing learning outcomes when compared with traditional lecture-based methods. Therefore, universities should consider integrating computer-based laboratories into their practical chemistry curricula to provide students with interactive and engaging learning experiences, especially when physical laboratories are inaccessible.

PMID:39546770 | DOI:10.2196/64476

Am J Manag Care. 2024 Nov;30(11):584-588. doi: 10.37765/ajmc.2024.89627.

ABSTRACT

OBJECTIVES: Provider directory inaccuracies have important implications for care navigation and access as well as ongoing regulatory efforts. We assessed the extent to which identified provider directory inaccuracies persisted across 7 specialties (cardiology, dermatology, endocrinology, gastroenterology, neurology, obstetrics-gynecology, primary care) and 5 carriers in the Pennsylvania Affordable Care Act insurance marketplace.

STUDY DESIGN: A secret shopper survey recontacted inaccurately listed providers (N = 1802) between 403 and 574 days after they were identified in an earlier secret shopper survey.

METHODS: Descriptive analyses, with tests of proportion and t tests to assess whether differences across carriers, specialties, and geographic locations were statistically significant.

RESULTS: Of 1802 inaccurate provider listings, 451 (25.0%) had been removed at follow-up, 966 providers (53.6%) were successfully contacted, and 385 providers (21.4%) could not be reached. Of the recontacted providers, 240 (13.3%) were listed accurately at follow-up and 726 (40.3%) were listed with various inaccuracies, including 31.0% (n = 558) with inaccurate contact information, 11.2% (n = 201) listed under the wrong specialty, and 1.9% (n = 34) erroneously listed as being in network despite being out of network. We found substantial differences across carriers and specialties but not by rurality. Inaccuracies also were less likely to persist in the state’s 2 metropolitan areas. Among inaccurate provider listings, on average, 540 days (median, 544 days) had passed between the initial and subsequent contacts.

CONCLUSIONS: A large number of provider directory inaccuracies persist well beyond the 90-day expectation mandated by federal regulations, raising substantial concerns about compliance. These inaccuracies may impose substantial barriers to patient access and may render existing assessments of network adequacy ineffective.

PMID:39546760 | DOI:10.37765/ajmc.2024.89627

Am J Manag Care. 2024 Nov;30(11):548-554. doi: 10.37765/ajmc.2024.89625.

ABSTRACT

OBJECTIVES: We assessed whether proactive care management for artificial intelligence (AI)-identified at-risk patients reduced preventable emergency department (ED) visits and hospital admissions (HAs).

STUDY DESIGN: Stepped-wedge cluster randomized design.

METHODS: Adults receiving primary care at 48 UCLA Health clinics and determined to be at risk based on a homegrown AI model were included. We employed a stepped-wedge cluster randomized design, assigning groups of clinics (pods) to 1 of 4 single-cohort waves during which the proactive care intervention was implemented. The primary end points were potentially preventable HAs and ED visits; secondary end points were all HAs and ED visits. Within each wave, we used an interrupted time series and segmented regression analysis to compare utilization trends.

RESULTS: In the pooled analysis of high-risk and highest-risk patients (n = 3007), potentially preventable HAs showed a statistically significant level drop (-27% [95% CI, -44% to -6%]), without any corresponding change in trends. Potentially preventable ED visits did not show a substantial level drop in response to the intervention, although a nonsignificant differential change in trend was observed, with visit rates decelerating 7% faster in the intervention cohorts (95% CI, -13% to 0%). Nonsignificant drops were observed for all HAs (-19% [95% CI, -35% to 1%]; P = .06) and ED visits (-15% [95% CI, -28% to 1%]; P = .06).

CONCLUSIONS: A care management intervention targeting AI-identified at-risk patients was followed by a onetime, significant, sizable reduction in preventable HA rates. Further exploration is needed to assess the potential of integrating AI and care management in preventing acute hospital encounters.

PMID:39546757 | DOI:10.37765/ajmc.2024.89625