Data quality in a survey of registered medical cannabis users with cancer: nonresponse and measurement error.

Cannabis use among individuals with cancer is best understood using survey self-report. As cannabis remains federally illegal, surveys could be subject to nonresponse and measurement issues impacting data quality. We surveyed individuals using medical cannabis for a cancer-related condition in the Minnesota Medical Cannabis Program (MCP). Although survey responders are older, there are no differences by race and ethnicity, gender, or receipt of reduced cannabis registry enrollment fee. Responders made a more recent purchase and more recently completed an independent symptom assessment for the registry than nonresponders, suggesting some opportunity for nonresponse error. Among responders, self-report and MCP administrative data with respect to age, race, gender, registry certification, and cannabis purchase history were similar. Responders were less likely to report receipt of Medicaid than would be expected based on registry low-income enrollment eligibility. Although attention should be paid to potential for nonresponse error, surveys are a reliable tool to ascertain cannabis behavior patterns in this population.

[The quality of the anthropometric data on children under five years of age in information systems in the State of São Paulo, Brazil]. / Qualidade dos dados antropométricos de crianças menores de cinco anos em sistemas de informação no estado de São Paulo, Brasil.

This paper involves the analysis of the quality of anthropometric data on children under five years of age in two information systems in the State of São Paulo. The sample included 2,117,108 children from the Food and Nutrition Surveillance System (SISVAN), and 748,551 from the State Milk Project (VIVALEITE). Initially, we evaluated the frequency of missing values and others outside the equipment spectrum and calculated the digit-to-weight preference index. After calculating height-for-age (HAZ), weight-for-age (WAZ), and body mass index-for-age (BAZ), we flagged the biologically implausible values (BIV) and calculated the standard deviation (SD). For each municipality, we calculated the mean and the SD of HAZ, WAZ, and BAZ; and plotted the SD values as a function of the mean. The digit-to-weight preference index was greater among children aged between 24 and 59 months in SISVAN. The frequency of BIV for HAZ (SISVAN 2.56%; VIVALEITE 0.98%) was higher than for WAZ (SISVAN 2.10%; VIVALEITE 0.18%). For HAZ, variations among municipalities were more pronounced in VIVALEITE than in SISVAN. The height variable presents low reliability in both systems. The weight variable reveals satisfactory quality in VIVALEITE and unsatisfactory quality in SISVAN.

O objetivo foi analisar a qualidade dos dados antropométricos de crianças menores de cinco anos em dois sistemas de informação no estado de São Paulo. A amostra compreendeu 2.117.108 crianças do Sistema de Vigilância Alimentar e Nutricional (Sisvan) e 748.551 do Projeto Estadual do Leite (Vivaleite). Inicialmente, avaliamos a frequência de valores faltantes e fora do espectro do equipamento, e calculamos o índice de preferência de dígito para peso. Após calcular os índices de altura para idade (A-I), peso para idade (P-I) e índice de massa corporal para idade (IMC-I), identificamos os valores biologicamente implausíveis (VBI) e calculamos o desvio-padrão (DP). Para cada município, calculamos a média e o DP de A-I, P-I e IMC-I; e plotamos os valores de DP em função da média. A preferência de dígito no peso foi maior em crianças de 24 a 59 meses no Sisvan. A frequência de VBI para A-I (SISVAN 2,56%; Vivaleite 0,98%) foi maior do que para P-I (Sisvan 2,10%; Vivaleite 0,18%). Para o índice A-I as variações entre os municípios foram mais acentuadas no Vivaleite do que no Sisvan. A variável altura apresentou baixa confiabilidade nos dois sistemas. A variável peso apresentou qualidade satisfatória no Vivaleite e insatisfatória no Sisvan.

Assessment of expanded programme on immunization routine data quality in the upper east region of Ghana.

BACKGROUND: Data quality is a major challenge for most health institutions and organizations across the globe. The Ghana Health Service, supported by other non-governmental organizations, has instituted various strategies to address and improve data quality issues in regional and district health facilities in Ghana. This study sought to assess routine data quality of Expanded Programme on Immunization, specifically for Penta 1 and Penta 3 vaccines. METHODS: A descriptive cross-sectional study design was used for the study. A simple random sampling method was used to select thirty-four health facilities across seven sub-municipalities. Records from the Expanded Programme on Immunization (EPI) Tally Books and Monthly Vaccination Summary Report were reviewed and compared with data entered into the District Health Information Management System 2 (DHIMS2) software for the period of January to December 2020. The World Health Organization Data quality self-assessment (DQS) tool was used to compare data recorded in the EPI tally books with monthly data from summary reports and DHIMS2. Data accuracy ratio was determined by the data quality assessment tools and STATA version 14.2 was used to run additional analysis. A data discrepancy is when two corresponding data sets don't match. RESULTS: The results showed discrepancies between recounted tallies in EPI tally books and summary reports submitted as well as DHIMS2. Verification factor of 97.4% and 99.3% and a discrepancy rate of 2.6 and 0.7 for Penta 1 and Penta 3 respectively were recorded for tallied data and summary reports. A verification factor of 100.5% and 99.9% and a discrepancy of -0.5 and 0.1 respectively for the same antigens were obtained for the summary reports and DHIMS2. Data timeliness was 90.7% and completeness was 100% for both antigens. CONCLUSION: The accuracy of Penta 1 and Penta 3 data on EPI in the Upper East Region of Ghana was high. The data availability, timeliness and completeness were also high.

Challenges in the evaluation of suicide prevention measures and quality of suicide data in Germany, Austria, and Switzerland: findings from qualitative expert interviews.

BACKGROUND: Suicide prevention requires diverse, integrated, and evidence-based measures. Comprehensive evaluation of interventions and reliable suicide data are crucial for guiding policy-making and advancing suicide prevention efforts. This study aimed to analyze current issues and gaps in the evaluation of suicide prevention measures and the quality of suicide data in Germany, Austria, and Switzerland to derive specific recommendations for improvement. METHODS: Online, semi-structured interviews were conducted with 36 experts in suicide prevention from Germany, Austria, and Switzerland, covering insights from policy, science, and practice. The interviews took place between September 2022 and February 2023, were audio-recorded, transcribed verbatim, and analyzed using the Framework method. RESULTS: While solid evidence supports the effectiveness of some suicide prevention interventions, experts indicated that the evaluation of many other measures is weak. Conducting effectiveness studies in suicide prevention presents a range of methodological and practical challenges, including recruitment difficulties, choosing adequate outcome criteria, ethical considerations, and trade-offs in allocating resources to evaluation efforts. Many interviewees rated the quality of national suicide statistics in Germany, Austria, and Switzerland as comparatively high. However, they noted limitations in the scope, timeliness, and reliability of these data, prompting some regions to implement their own suicide monitoring systems. None of the three countries has national routine data on suicide attempts. CONCLUSION: While some challenges in evaluating suicide prevention measures are inevitable, others can potentially be mitigated. Evaluations could be enhanced by combining traditional and innovative research designs, including intermediate outcomes and factors concerning the implementation process, and employing participatory and transdisciplinary research to engage different stakeholders. Reliable suicide data are essential for identifying trends, supporting research, and designing targeted prevention measures. To improve the quality of suicide data, a standardized monitoring approach, including uniform definitions, trained professionals, and cross-sector agreement on leadership and financing, should be pursued. This study provides actionable recommendations and highlights existing good practice approaches, thereby supporting decision-makers and providing guidance for advancing suicide prevention on a broader scale.

A standardized image processing and data quality platform for rodent fMRI.

Functional magnetic resonance imaging in rodents holds great potential for advancing our understanding of brain networks. Unlike the human community, there remains no standardized resource in rodents for image processing, analysis and quality control, posing significant reproducibility limitations. Our software platform, Rodent Automated Bold Improvement of EPI Sequences, is a pipeline designed to address these limitations for preprocessing, quality control, and confound correction, along with best practices for reproducibility and transparency. We demonstrate the robustness of the preprocessing workflow by validating performance across multiple acquisition sites and both mouse and rat data. Building upon a thorough investigation into data quality metrics across acquisition sites, we introduce guidelines for the quality control of network analysis and offer recommendations for addressing issues. Taken together, this software platform will allow the emerging community to adopt reproducible practices and foster progress in translational neuroscience.

Race adjustments in clinical algorithms can help correct for racial disparities in data quality.

Despite ethical and historical arguments for removing race from clinical algorithms, the consequences of removal remain unclear. Here, we highlight a largely undiscussed consideration in this debate: varying data quality of input features across race groups. For example, family history of cancer is an essential predictor in cancer risk prediction algorithms but is less reliably documented for Black participants and may therefore be less predictive of cancer outcomes. Using data from the Southern Community Cohort Study, we assessed whether race adjustments could allow risk prediction models to capture varying data quality by race, focusing on colorectal cancer risk prediction. We analyzed 77,836 adults with no history of colorectal cancer at baseline. The predictive value of self-reported family history was greater for White participants than for Black participants. We compared two cancer risk prediction algorithms-a race-blind algorithm which included standard colorectal cancer risk factors but not race, and a race-adjusted algorithm which additionally included race. Relative to the race-blind algorithm, the race-adjusted algorithm improved predictive performance, as measured by goodness of fit in a likelihood ratio test (P-value: <0.001) and area under the receiving operating characteristic curve among Black participants (P-value: 0.006). Because the race-blind algorithm underpredicted risk for Black participants, the race-adjusted algorithm increased the fraction of Black participants among the predicted high-risk group, potentially increasing access to screening. More broadly, this study shows that race adjustments may be beneficial when the data quality of key predictors in clinical algorithms differs by race group.

Assessing the quality of emergency department data for injury surveillance in Victoria, Australia: a comparative analysis of two Victorian hospital data sources.

OBJECTIVE: The emergency department (ED) is pivotal in treating serious injuries, making it a valuable source for population-based injury surveillance. In Victoria, information that is relevant to injury surveillance is collected in the Victorian Emergency Minimum Dataset (VEMD). This study aims to assess the data quality of the VEMD as an injury data source by comparing it with the Victorian Admitted Episodes Dataset (VAED). DESIGN: A retrospective observational study of administrative healthcare data. SETTING AND PARTICIPANTS: VEMD and VAED data from July 2014 to June 2019 were compared. Including only hospitals contributing to both datasets, cases that (1) arrived at the ED and (2) were subsequently admitted, were selected. RESULTS: While the overall number of cases was similar, VAED outnumbered VEMD cases (414 630 vs 404 608), suggesting potential under-reporting of injuries in the ED. Age-related differences indicated a relative under-representation of older individuals in the VEMD. Injuries caused by falls or transport, and intentional injuries were relatively under-reported in the VEMD. CONCLUSIONS: Injury cases were more numerous in the VAED than in the VEMD even though the number is expected to be equal based on case selection. Older patients were under-represented in the VEMD; this could partly be attributed to patients being admitted for an injury after they presented to the ED with a non-injury ailment. The patterns of under-representation described in this study should be taken into account in ED-based injury incidence reporting.

The National Swedish Lymphoma Register - a systematic validation of data quality.

BACKGROUND AND PURPOSE: The Swedish Lymphoma Register (SLR) was initiated in the year 2000 with the aim to monitor quality of care in diagnostics, treatment and outcome of all lymphomas diagnosed nationally among adults. Here, we present the first systematic validation of SLR records as a basis for improved register quality and patient care. PATIENTS AND METHODS: We evaluated timeliness and completeness of register records among patients diagnosed with lymphoma in the SLR (n = 16,905) compared with the National Cancer Register for the period 2013-2020. Comparability was assessed through evaluation of coding routines against national and international guidelines. Accuracy of 42 variables was evaluated through re-abstraction of data from medical records among 600 randomly selected patients diagnosed in 2016-2017 and treated across all six Swedish healthcare regions.  Results: Completeness was high, >95% per year for the period 2013-2018, and >89% for 2019-2020 compared to the National Cancer Register. One in four patients was registered within 3 months, and 89.9% within 2 years of diagnosis. Registration instructions and coding procedures followed the prespecified guidelines. Missingness was generally low (<5%), but high for occasional variables, for example, those describing maintenance and consolidative treatment. Exact agreement of categorical variables was high overall (>80% for 24/34 variables), especially for treatment-related data (>80% for 17/19 variables). INTERPRETATION: Completeness and accuracy are high in the SLR, while timeliness could be improved. Finetuning of variable registration guided by this validation can further improve reliability of register reports and advance service to lymphoma patients and health care in the future.

The trend of the quality of cause-of-death data and its association with socio-economic indicators in Serbia in the period 2005-19.

This study aims to evaluate the temporal trend in the quality of cause-of-death data and garbage code profiles and to determine its association with socio-economic status in Serbia. A longitudinal study was assessed using data from mortality registers from 2005 to 2019. Computer application Analysis of Causes of National Deaths for Action (ANACONDA) calculates the distribution of garbage codes by severity and composite quality indicator: Vital Statistics Performance Index for Quality (VSPI(Q)). A relationship between VSPI(Q) and country development was estimated by analysing two socio-economic indicators: the Socio-demographic Index and the Human Development Index (HDI). Serbia indicates progress in strengthening cause-of-death statistics. The steady upward trend of the VSPI(Q) index has risen from 55.6 (medium quality) to 70.2 (high quality) over the examined years. Significant reduction of 'Insufficiently specified causes with limited impact' (Level 4) and an increase in the trend of 'High-impact garbage codes' (Levels 1 to 3) were evident. Decreased deaths of no policy value (annual percentage change of -1.41%) have manifested since 2014. A strong positive association between VSPI(Q) and socio-economic indicators was assessed, where the HDI has shown a stronger association with VSPI(Q). Improved socio-economic conditions on the national level are followed by enhanced cause-of-death data quality. Upcoming actions to improve quality should be directed at high-impact garbage codes. The study underlines the need to prioritise the education and training of physicians with a crucial role in death certification to overcome many cause-of-death quality issues identified in this assessment.

Data quality assessment of the Enhanced Gonococcal Antimicrobial Surveillance Programme (EGASP), Thailand, 2015-2021.

BACKGROUND: Quality assessments of gonococcal surveillance data are critical to improve data validity and to enhance the value of surveillance findings. Detecting data errors by systematic audits identifies areas for quality improvement. We designed and implemented an internal audit process to evaluate the accuracy and completeness of surveillance data for the Thailand Enhanced Gonococcal Antimicrobial Surveillance Programme (EGASP). METHODS: We conducted a data quality audit of source records by comparison with the data stored in the EGASP database for five audit cycles from 2015-2021. Ten percent of culture-confirmed cases of Neisseria gonorrhoeae were randomly sampled along with any cases identified with elevated antimicrobial susceptibility testing results and cases with repeat infections. Incorrect and incomplete data were investigated, and corrective action and preventive actions (CAPA) were implemented. Accuracy was defined as the percentage of identical data in both the source records and the database. Completeness was defined as the percentage of non-missing data from either the source document or the database. Statistical analyses were performed using the t-test and the Fisher's exact test. RESULTS: We sampled and reviewed 70, 162, 85, 68, and 46 EGASP records during the five audit cycles. Overall accuracy and completeness in the five audit cycles ranged from 93.6% to 99.4% and 95.0% to 99.9%, respectively. Overall, completeness was significantly higher than accuracy (p = 0.017). For each laboratory and clinical data element, concordance was >85% in all audit cycles except for two laboratory data elements in two audit cycles. These elements significantly improved following identification and CAPA implementation. DISCUSSION: We found a high level of data accuracy and completeness in the five audit cycles. The implementation of the audit process identified areas for improvement. Systematic quality assessments of laboratory and clinical data ensure high quality EGASP surveillance data to monitor for antimicrobial resistant Neisseria gonorrhoeae in Thailand.

Data quality and timeliness analysis for post-vaccination adverse event cases reported through healthcare data exchange to FDA BEST pilot platform.

Introduction: This study is part of the U.S. Food and Drug Administration (FDA)'s Biologics Effectiveness and Safety (BEST) initiative, which aims to improve the FDA's postmarket surveillance capabilities by using real-world data (RWD). In the United States, using RWD for postmarket surveillance has been hindered by the inability to exchange clinical data between healthcare providers and public health organizations in an interoperable format. However, the Office of the National Coordinator for Health Information Technology (ONC) has recently enacted regulation requiring all healthcare providers to support seamless access, exchange, and use of electronic health information through the interoperable HL7 Fast Healthcare Interoperability Resources (FHIR) standard. To leverage the recent ONC changes, BEST designed a pilot platform to query and receive the clinical information necessary to analyze suspected AEs. This study assessed the feasibility of using the RWD received through the data exchange of FHIR resources to study post-vaccination AE cases by evaluating the data volume, query response time, and data quality. Materials and methods: The study used RWD from 283 post-vaccination AE cases, which were received through the platform. We used descriptive statistics to report results and apply 322 data quality tests based on a data quality framework for EHR. Results: The volume analysis indicated the average clinical resources for a post-vaccination AE case was 983.9 for the median partner. The query response time analysis indicated that cases could be received by the platform at a median of 3 min and 30 s. The quality analysis indicated that most of the data elements and conformance requirements useful for postmarket surveillance were met. Discussion: This study describes the platform's data volume, data query response time, and data quality results from the queried postvaccination adverse event cases and identified updates to current standards to close data quality gaps.

[Managing routine health information system data during a security crisis in the health district of Timbuktu, Mali in 2023]. / La gestion des données du système d'information sanitaire de routine en période de crise sécuritaire dans le district sanitaire de Tombouctou, au Mali en 2023.

Introduction: an effective health information system (HIS) ensures the production, analysis, dissemination and use of reliable and up-to-date information on the determinants of health. However, it can encounter obstacles that hinder its functioning, such as armed conflicts, which limit access and quality of healthcare services. The purpose of our study was to help improve data management for routine health information system in the health district of Timbuktu during a security crisis. Methods: we conducted a descriptive cross-sectional study, among health information management professionals in the Timbuktu Health District from 15 April to 08 September 2023. Data obtained from a survey questionnaire were analyzed using Epi Info version 7.2.2. and processed using Microsoft Word and Excel 2016. Results: a total of 6 health facilities were surveyed. Data collection, analysis and feedback were very poor. Data quality was 100% complete, 92.40% prompt and 68.11% accurate. The major constraints were: low involvement of health workers in the SIS (22.22%), insufficient training on the SISR (29.63%), supervision (47.06%), internet inaccessibility (66.67%), feeling of insecurity (37.04%) and fear (61.76%) in health facilities. Conclusion: our results show low-level processes, poor network coverage, shortage of qualified health information management professionals and increasing insecurity. A broader mixed-methods research would provide a better understanding.

Data quality in Brazilian population-based cancer registries for gastrointestinal cancers.

BACKGROUND: Population-based cancer registries (PBCRs) are the primary source of information for cancer surveillance and monitoring. Currently, there are 30 active PBCRs in Brazil. The objective of this study was to analyze the data quality of five gastrointestinal cancers (esophagus, stomach, colorectal, liver, and pancreas) according to the criteria of comparability, validity, completeness, and timeliness in Brazilian cancer registries. METHODS: This study included data from Brazilian PBCRs with more than ten years of historical data starting in the year 2000, regardless of the type of defined geographical coverage (state, metropolitan region, or capital), totaling 16 registries. Brazilian PBCRs were evaluated based on four international data quality criteria: comparability, validity (accuracy), completeness, and timeliness. All cancer cases were analyzed, except for nonmelanoma skin cancer cases (C44) and five gastrointestinal tumors (esophageal cancer, stomach cancer, colorectal cancer, liver cancer, and pancreatic cancer) per cancer registry and sex, according to the available period. RESULTS: The 16 Brazilian PBCRs represent 17% of the population (36 million inhabitants in 2021) according to data from 2000 to 2018. There was a variation in the incidence in the historical series ranging from 12 to 19 years. The proportion of morphologically verified (MV%) cases varied from 74.3% (Manaus) to 94.8% (Aracaju), and the proportion of incidentally reported death certificate only (DCO%) cases varied from 3.0% (São Paulo) to 23.9% (Espírito Santo). High-lethality malignant neoplasms, such as liver and pancreas, had DCO percentages greater than 30% in most cancer registries. The sixteen registries have more than a 48-month delay in data release compared to the 2022 calendar year. CONCLUSION: The studied Brazilian cancer registries met international comparability criteria; however, half of the registries showed indices below the expected levels for validity and completeness criteria for high-lethality tumors such as liver and pancreas tumors, in addition to a long delay in data availability and disclosure. Significant efforts are necessary to ensure the operational and stability of the PBCR in Brazil, which continues to be a tool for monitoring cancer incidence and assessing national cancer control policies.

Evaluation of Errors on Death Certificates.

BACKGROUND: Death certificates provide vital data for disease surveillance and health policy. However, errors are common globally, undermining data reliability. This study analyzed prevalence and types of errors in death certificates at a tertiary hospital in Nepal. METHODS: A cross-sectional study reviewed all death certificates issued at Lumbini Medical College, Nepal from April 2020 to April 2022. Certificates were assessed for errors including improper sequencing, absent time intervals, abbreviations, illegible writing, and inaccurate immediate, antecedent, and underlying causes of death as per international guidelines. Errors were classified as major or minor. RESULTS: Of 139 certificates, none were error-free. The most common error was incorrectly or incompletely filling the immediate cause of death (77.7%). Other errors included absent time of death (17.3%), abbreviations (57.6%), illegible writing (22.3%), and omitting the hospital stamp/medical council registration number (8.6%). Based on international criteria, 76.3% had minor errors, 23% had both major and minor errors. CONCLUSIONS: This study found a high rate of errors in death certification at a tertiary hospital in Nepal, undermining data accuracy. Regular training and monitoring with feedback are recommended to improve certification practices. Accurate cause-of-death data is vital for healthcare policy and decision-making in Nepal.

Mitigating data quality challenges in ambulatory wrist-worn wearable monitoring through analytical and practical approaches.

Chronic disease management and follow-up are vital for realizing sustained patient well-being and optimal health outcomes. Recent advancements in wearable technologies, particularly wrist-worn devices, offer promising solutions for longitudinal patient monitoring, replacing subjective, intermittent self-reporting with objective, continuous monitoring. However, collecting and analyzing data from wearables presents several challenges, such as data entry errors, non-wear periods, missing data, and wearable artifacts. In this work, we explore these data analysis challenges using two real-world datasets (mBrain21 and ETRI lifelog2020). We introduce practical countermeasures, including participant compliance visualizations, interaction-triggered questionnaires to assess personal bias, and an optimized pipeline for detecting non-wear periods. Additionally, we propose a visualization-oriented approach to validate processing pipelines using scalable tools such as tsflex and Plotly-Resampler. Lastly, we present a bootstrapping methodology to evaluate the variability of wearable-derived features in the presence of partially missing data segments. Prioritizing transparency and reproducibility, we provide open access to our detailed code examples, facilitating adaptation in future wearable research. In conclusion, our contributions provide actionable approaches for improving wearable data collection and analysis.

Factors Influencing Data Quality in Electronic Health Record Systems in 50 Health Facilities in Rwanda and the Role of Clinical Alerts: Cross-Sectional Observational Study.

BACKGROUND: Electronic health records (EHRs) play an increasingly important role in delivering HIV care in low- and middle-income countries. The data collected are used for direct clinical care, quality improvement, program monitoring, public health interventions, and research. Despite widespread EHR use for HIV care in African countries, challenges remain, especially in collecting high-quality data. OBJECTIVE: We aimed to assess data completeness, accuracy, and timeliness compared to paper-based records, and factors influencing data quality in a large-scale EHR deployment in Rwanda. METHODS: We randomly selected 50 health facilities (HFs) using OpenMRS, an EHR system that supports HIV care in Rwanda, and performed a data quality evaluation. All HFs were part of a larger randomized controlled trial, with 25 HFs receiving an enhanced EHR with clinical decision support systems. Trained data collectors visited the 50 HFs to collect 28 variables from the paper charts and the EHR system using the Open Data Kit app. We measured data completeness, timeliness, and the degree of matching of the data in paper and EHR records, and calculated concordance scores. Factors potentially affecting data quality were drawn from a previous survey of users in the 50 HFs. RESULTS: We randomly selected 3467 patient records, reviewing both paper and EHR copies (194,152 total data items). Data completeness was >85% threshold for all data elements except viral load (VL) results, second-line, and third-line drug regimens. Matching scores for data values were close to or >85% threshold, except for dates, particularly for drug pickups and VL. The mean data concordance was 10.2 (SD 1.28) for 15 (68%) variables. HF and user factors (eg, years of EHR use, technology experience, EHR availability and uptime, and intervention status) were tested for correlation with data quality measures. EHR system availability and uptime was positively correlated with concordance, whereas users' experience with technology was negatively correlated with concordance. The alerts for missing VL results implemented at 11 intervention HFs showed clear evidence of improving timeliness and completeness of initially low matching of VL results in the EHRs and paper records (11.9%-26.7%; P<.001). Similar effects were seen on the completeness of the recording of medication pickups (18.7%-32.6%; P<.001). CONCLUSIONS: The EHR records in the 50 HFs generally had high levels of completeness except for VL results. Matching results were close to or >85% threshold for nondate variables. Higher EHR stability and uptime, and alerts for entering VL both strongly improved data quality. Most data were considered fit for purpose, but more regular data quality assessments, training, and technical improvements in EHR forms, data reports, and alerts are recommended. The application of quality improvement techniques described in this study should benefit a wide range of HFs and data uses for clinical care, public health, and disease surveillance.

Better performance of deep learning pulmonary nodule detection using chest radiography with pixel level labels in reference to computed tomography: data quality matters.

Labeling errors can significantly impact the performance of deep learning models used for screening chest radiographs. The deep learning model for detecting pulmonary nodules is particularly vulnerable to such errors, mainly because normal chest radiographs and those with nodules obscured by ribs appear similar. Thus, high-quality datasets referred to chest computed tomography (CT) are required to prevent the misclassification of nodular chest radiographs as normal. From this perspective, a deep learning strategy employing chest radiography data with pixel-level annotations referencing chest CT scans may improve nodule detection and localization compared to image-level labels. We trained models using a National Institute of Health chest radiograph-based labeling dataset and an AI-HUB CT-based labeling dataset, employing DenseNet architecture with squeeze-and-excitation blocks. We developed four models to assess whether CT versus chest radiography and pixel-level versus image-level labeling would improve the deep learning model's performance to detect nodules. The models' performance was evaluated using two external validation datasets. The AI-HUB dataset with image-level labeling outperformed the NIH dataset (AUC 0.88 vs 0.71 and 0.78 vs. 0.73 in two external datasets, respectively; both p < 0.001). However, the AI-HUB data annotated at the pixel level produced the best model (AUC 0.91 and 0.86 in external datasets), and in terms of nodule localization, it significantly outperformed models trained with image-level annotation data, with a Dice coefficient ranging from 0.36 to 0.58. Our findings underscore the importance of accurately labeled data in developing reliable deep learning algorithms for nodule detection in chest radiography.

Characterizing physician directory data quality: variation by specialty, state, and insurer.

BACKGROUND: As U.S. legislators are urged to combat ghost networks in behavioral health and address the provider data quality issue, it becomes important to better characterize the variation in data quality of provider directories to understand root causes and devise solutions. Therefore, this manuscript examines consistency of address, phone number, and specialty information for physician entries from 5 national health plan provider directories by insurer, physician specialty, and state. METHODS: We included all physicians in the Medicare Provider Enrollment, Chain, and Ownership System (PECOS) found in ≥ 2 health insurer physician directories across 5 large national U.S. health insurers. We examined variation in consistency of address, phone number, and specialty information among physicians by insurer, physician specialty, and state. RESULTS: Of 634,914 unique physicians in the PECOS database, 449,282 were found in ≥ 2 directories and included in our sample. Across insurers, consistency of address information varied from 16.5 to 27.9%, consistency of phone number information varied from 16.0 to 27.4%, and consistency of specialty information varied from 64.2 to 68.0%. General practice, family medicine, plastic surgery, and dermatology physicians had the highest consistency of addresses (37-42%) and phone numbers (37-43%), whereas anesthesiology, nuclear medicine, radiology, and emergency medicine had the lowest consistency of addresses (11-21%) and phone numbers (9-14%) across health insurer directories. There was marked variation in consistency of address, phone number, and specialty information by state. CONCLUSIONS: In evaluating a large national sample of U.S. physicians, we found minimal variation in provider directory consistency by insurer, suggesting that this is a systemic problem that insurers have not solved, and considerable variation by physician specialty with higher quality data among more patient-facing specialties, suggesting that physicians may respond to incentives to improve data quality. These data highlight the importance of novel policy solutions that leverage technology targeting data quality to centralize provider directories so as not to not reinforce existing data quality issues or policy solutions to create national and state-level standards that target both insurers and physician groups to maximize quality of provider information.

Development of Data Quality Indicators for Improving Hospital International Classification of Diseases-Coded Health Data Quality Globally.

BACKGROUND: Hospital inpatient data, coded using the International Classification of Diseases (ICD), is widely used to monitor diseases, allocate resources and funding, and evaluate patient outcomes. As such, hospital data quality should be measured before use; however, currently, there is no standard and international approach to assess ICD-coded data quality. OBJECTIVE: To develop a standardized method for assessing hospital ICD-coded data quality that could be applied across countries: Data quality indicators (DQIs). RESEARCH DESIGN: To identify a set of candidate DQIs, we performed an environmental scan, reviewing gray and academic literature on data quality frameworks and existing methods to assess data quality. Indicators from the literature were then appraised and selected through a 3-round Delphi process. The first round involved face-to-face group and individual meetings for idea generation, while the second and third rounds were conducted remotely to collect online ratings. Final DQIs were selected based on the panelists' quantitative and qualitative feedback. SUBJECTS: Participants included international experts with expertise in administrative health data, data quality, and ICD coding. RESULTS: The resulting 24 DQIs encompass 5 dimensions of data quality: relevance, accuracy and reliability; comparability and coherence; timeliness; and Accessibility and clarity. These will help stakeholders (eg, World Health Organization) to assess hospital data quality using the same standard across countries and highlight areas in need of improvement. CONCLUSIONS: This novel area of research will facilitate international comparisons of ICD-coded data quality and be valuable to future studies and initiatives aimed at improving hospital administrative data quality.

Effectiveness of remote risk-based monitoring and potential benefits for combination with direct data capture.

BACKGROUND: In recent years, alternative monitoring approaches, such as risk-based and remote monitoring techniques, have been recommended instead of traditional on-site monitoring to achieve more efficient monitoring. Remote risk-based monitoring (R2BM) is a monitoring technique that combines risk-based and remote monitoring and focuses on the detection of critical data and process errors. Direct data capture (DDC), which directly collects electronic source data, can facilitate R2BM by minimizing the extent of source documents that must be reviewed and reducing the additional workload on R2BM. In this study, we evaluated the effectiveness of R2BM and the synergistic effect of combining R2BM with DDC. METHODS: R2BM was prospectively conducted with eight participants in a randomized clinical trial using a remote monitoring system that uploaded photographs of source documents to a cloud location. Critical data and processes were verified by R2BM, and later, all were confirmed by on-site monitoring to evaluate the ability of R2BM to detect critical data and process errors and the workload of uploading photographs for clinical trial staff. In addition, the reduction of the number of uploaded photographs was evaluated by assuming that the DDC was introduced for data collection. RESULTS: Of the 4645 data points, 20.9% (n = 973, 95% confidence interval = 19.8-22.2) were identified as critical. All critical data errors corresponding to 5.4% (n = 53/973, 95% confidence interval = 4.1-7.1) of the critical data and critical process errors were detectable by R2BM. The mean number of uploaded photographs and the mean time to upload them per visit per participant were 34.4 ± 11.9 and 26.5 ± 11.8 min (mean ± standard deviation), respectively. When assuming that DDC was introduced for data collection, 45.0% (95% confidence interval = 42.2-47.9) of uploaded photographs for R2BM were reduced. CONCLUSIONS: R2BM can detect 100% of the critical data and process errors without on-site monitoring. Combining R2BM with DDC reduces the workload of R2BM and further improves its efficiency.