Missing Codes Research Articles

A retrospective analysis using comorbidity detecting algorithmic software to determine the incidence of International Classification of Diseases (ICD) code omissions and appropriateness of Diagnosis-Related Group (DRG) code modifiers

BackgroundThe mechanism for recording International Classification of Diseases (ICD) and diagnosis related groups (DRG) codes in a patient’s chart is through a certified medical coder who manually reviews the medical record at the completion of an admission. High-acuity ICD codes justify DRG modifiers, indicating the need for escalated hospital resources. In this manuscript, we demonstrate that value of rules-based computer algorithms that audit for omission of administrative codes and quantifying the downstream effects with regard to financial impacts and demographic findings did not indicate significant disparities.MethodsAll study data were acquired via the UCLA Department of Anesthesiology and Perioperative Medicine’s Perioperative Data Warehouse. The DataMart is a structured reporting schema that contains all the relevant clinical data entered into the EPIC (EPIC Systems, Verona, WI) electronic health record. Computer algorithms were created for eighteen disease states that met criteria for DRG modifiers. Each algorithm was run against all hospital admissions with completed billing from 2019. The algorithms scanned for the existence of disease, appropriate ICD coding, and DRG modifier appropriateness. Secondarily, the potential financial impact of ICD omissions was estimated by payor class and an analysis of ICD miscoding was done by ethnicity, sex, age, and financial class.ResultsData from 34,104 hospital admissions were analyzed from January 1, 2019, to December 31, 2019. 11,520 (32.9%) hospital admissions were algorithm positive for a disease state with no corresponding ICD code. 1,990 (5.8%) admissions were potentially eligible for DRG modification/upgrade with an estimated lost revenue of $22,680,584.50. ICD code omission rates compared against reference groups (private payors, Caucasians, middle-aged patients) demonstrated significant p-values < 0.05; similarly significant p-value where demonstrated when comparing patients of opposite sexes.ConclusionsWe successfully used rules-based algorithms and raw structured EHR data to identify omitted ICD codes from inpatient medical record claims. These missing ICD codes often had downstream effects such as inaccurate DRG modifiers and missed reimbursement. Embedding augmented intelligence into this problematic workflow has the potential for improvements in administrative data, but more importantly, improvements in administrative data accuracy and financial outcomes.

Expanding the time-interleaving design capabilities: A 28 GS/s 4-bit time-interleaved current-steering DAC case study

In this work, a novel approach of designing high-speed time-interleaved Digital-to-Analog Converters (DACs), that exploits high-order time-interleaved factors, was proposed. The presented time-interleaving design approach is based on the current superposition principle, capable of expanding the time-interleaved factor of DACs without compromising the conversion linearity and accuracy. For the validation of the proposed design approach, a 28 GS/s 4-bit 4 × Time-Interleaved current-steering DAC was designed using a 22 nm Fully-Depleted Silicon-On-Insulator (FDSOI) CMOS process node. Post-layout simulations were carried out by developing a custom, hybrid RC/RLCk parasitic extraction methodology, capable of capturing all possible layout parasitic effects due to the high conversion speed of the designed DAC. Using the proposed approach, the designed time-interleaved DAC was capable of achieving ENOB>3.83 bits, SFDR>28.7 dBc for fin≤1.75 GHz, with no missing codes and a low power consumption of Pdiss=3.1 mW/core.

Variations in Using Diagnosis Codes for Defining Age-Related Macular Degeneration Cohorts

Data harmonization is vital for secondary electronic health record data analysis, especially when combining data from multiple sources. Currently, there is a gap in knowledge as to how studies identify cohorts of patients with age-related macular degeneration (AMD), a leading cause of blindness. We hypothesize that there is variation in using medical condition codes to define cohorts of AMD patients that can lead to either the under- or overrepresentation of such cohorts. This study identified articles studying AMD using the International Classification of Diseases (ICD-9, ICD-9-CM, ICD-10, and ICD-10-CM). The data elements reviewed included the year of publication; dataset origin (Veterans Affairs, registry, national or commercial claims database, and institutional EHR); total number of subjects; and ICD codes used. A total of thirty-seven articles were reviewed. Six (16%) articles used cohort definitions from two ICD terminologies. The Medicare database was the most used dataset (14, 38%), and there was a noted increase in the use of other datasets in the last few years. We identified substantial variation in the use of ICD codes for AMD. For the studies that used ICD-10 terminologies, 7 (out of 9, 78%) defined the AMD codes correctly, whereas, for the studies that used ICD-9 and 9-CM terminologies, only 2 (out of 30, 7%) defined and utilized the appropriate AMD codes (p = 0.0001). Of the 43 cohort definitions used from 37 articles, 31 (72%) had missing or incomplete AMD codes used, and only 9 (21%) used the exact codes. Additionally, 13 articles (35%) captured ICD codes that were not within the scope of AMD diagnosis. Efforts to standardize data are needed to provide a reproducible research output.

The impact of SNOMED CT on clinical coding quality and productivity

Background/Aims Studies have shown that using the clinical terminology SNOMED CT can improve clinical documentation, data quality and workforce productivity. This study aimed to evaluate the impact of a SNOMED CT-based digital tool on the quality, completeness and efficiency of clinical coding. Methods A SNOMED CT-enabled clinical coding tool (SNOMED CT - Clinical Coding Assistant) was created. The tool used nursing assessments to identify relevant conditions that should be coded. SNOMED CT concepts and the nationally published SNOMED CT to ICD-10 cross-maps were compared to clinical coding data generated by the typical clinical coding and audit processes. The tool's impact on coding quality, completeness and productivity was assessed using various statistical measures. A total of 1681 coded hospital spells, representing 1589 patients, were included, across a 2-month period. Results Use of the SNOMED CT - Clinical Coding Assistant led to ICD-10 codes being added or changed in nearly half (n=773, 46.0%) of the hospital spells. The tool identified that over a third of the spells (n=588, 35.0%) had missing ICD-10 codes. The tool significantly reduced the time taken to find and validate the presence of numerous conditions for subsequent coding (P&lt;0.001). Conclusions Use of SNOMED CT-based tools can improve clinical coding accuracy, completeness and productivity. SNOMED CT has an important role to play in clinical coding transformation and effective system implementation is critical.

Planning the Migration to a New Database: Implications for the Collections of Meise Botanic Garden

Established in 1796, the Meise Botanic Garden has undergone multiple changes in nationality, name, focus, and locations. These changes have impacted its herbarium, living collections, and associated data. The way specimens are filed in the herbarium varies according to specimen type, taxonomic group and historical importance. Pragmatic solutions were used in the past for dealing with the divergent collections and their users' specific needs. There are at least 15 filing systems used at the Botanic Garden at present. In 2015, funding facilitated the digitization of 1.2 million herbarium specimens. A follow-on project in 2018 digitzed another 1.4 million specimens. Together with other initiatives, like the Global Plants Initiative, we have more than 2.7 million imaged specimens, representing about 77% of our herbarium collections. Of the imaged specimens, 89% have minimal data (barcode, species name, collector(s), collector number, collection date, country and locality), while the remaining 11% still need the metadata captured from the scanned images (mostly being done through the crowdsourcing project DoeDat). In 2023, a call to tender was started for a new Collection Management System (CMS) as part of the Distributed System of Scientific Collections (DiSSCo) Flanders*1 project. This call for tender was compiled in consultation with various natural history institutions, botanic gardens and museums in Belgium. At present, we are in the process of evaluating the various tenders. As part of the preparation to migrate the data from our present database to the new database, data gaps are being filled and data is being cleaned where possible. This will facilitate the migration from one system to another. This is particularly important for fields relating to taxonomy, countries and collectors, as these have implications for the physical location of specimens in the herbarium. In the past months, much time has been dedicated to linking specimens to the taxon table. At present fewer than 0.6% of specimens with a filing name (the scientific name under which the specimen is filed in the herbarium cupboards) remain to be linked to the taxon table. The use of the Taxonomic Name Resolution Service helped to resolve some of the issues associated with unlinked names. Many of the remaining unlinked names are names of infraspecific taxa (which are not yet in the taxon table), orthographic variations, or names of commercial cultivars or hybrids. However, this does not include the 11% of the specimens without data mentioned above. The number of specimens lacking a country code is 7%. The African Herbarium, which represents about a third of the specimens at the Botanic Garden, is sorted according to species, country and collector. Resolving missing country codes would facilitate the curatorial process for this collection. Likewise, 6% of specimen data only have collectors as free-text that still needs to be linked to the standardised collectors table. Atomizing and linking these collector strings would also ease the migration to a new CMS. In 2015, a master plan was also drawn up for the physical renovation of the Garden, including the herbarium's infrastructure. Work on the herbarium is planned for 2025, but is contingent on funding. The CMS will form a crucial part of this move as specimens will be reclassified. As is mentioned above, there are several classification systems used at the Botanic Garden. The African and Belgian collections currently follow Mabberley (1997) for angiosperms. This system is no longer up-to-date, and we are considering moving to APG IV (Angiosperm Phylogeny Group) system for flowering plants. For example, the 213 genera present at the Botanic Garden for the family Liliaceae sensu lato (Mabberley 1997) would resolve to 24 families in APG IV. As the specimens are arranged alphabetically by family and genus, this would mean a significant physical reorganization of specimens. The CMS will also be used to track the movement of the various collections. Once the new CMS has been chosen, we will have a better idea of what needs to be done in order to facilitate data migration. For example, linking our taxonomic names to external sources using their permanent identifiers (e.g., World Flora Online, International Plant Names Index, Global Biodiversity Information Facility, Index Fungorum, MycoBank, World Register of Marine Species) would allow us to track and maintain taxonomic changes more easily in the future. Importing the various permanent identifiers into our current database would be prone to inconsistencies as the existing fields are not fit for the purpose. A possible solution would be to manage this data externally and import them into the new CMS. A similar approach could be followed with collectors and Wikidata permanent identifiers. We hope to develop a pragmatic approach with practical solutions.

Optimising vitrectomy operation note coding with machine learning.

The accurate encoding of operation notes is essential for activity-based funding and workforce planning. The aim of this project was to evaluate the procedural coding accuracy of vitrectomy and to develop machine learning, natural language processing (NLP) models that may assist with this task. This retrospective cohort study involved vitrectomy operation notes between a 21-month period at the Royal Adelaide Hospital. Coding of procedures were based on the Medicare Benefits Schedule (MBS)-the Australian equivalent to the Current Procedural Terminology (CPT®) codes used in the United States. Manual encoding was conducted for all procedures and reviewed by two vitreoretinal consultants. XGBoost, random forest and logistic regression models were developed for classification experiments. A cost-based analysis was subsequently conducted. There were a total of 1724 procedures with individual codes performed within 617 vitrectomy operation notes totalling $1 528 086.60 after manual review. A total of 1147 (66.5%) codes were missed in the original coding that amounted to $736 539.20 (48.2%). Our XGBoost model had the highest classification accuracy (94.6%) in the multi-label classification for the five most common procedures. The XGBoost model was the most successful model in identifying operation notes with two or more missing codes with an AUC of 0.87 (95% CI 0.80-0.92). Machine learning has been successful in the classification of vitrectomy operation note encoding. We recommend a combined human and machine learning approach to clinical coding as automation may facilitate more accurate reimbursement and enable surgeons to prioritise higher quality clinical care.

Diagnostic groups of hospital stays and outpatient visits during 10 years before Alzheimer’s disease

BackgroundAlzheimer’s disease (AD) is a major determinant of healthcare costs and increase in the healthcare service use occur already before the AD diagnosis. However, little is known how the different diagnosis categories contribute to this increase in healthcare use. We investigated how the hospitalizations and specialized healthcare outpatient visits from different diagnosis categories, based on the International Classification of Diseases (ICD-10) chapters, contribute to increased specialized healthcare service use during ten-year period preceding AD diagnosis.MethodsA register-based nationwide cohort of 42,934 community-dwelling persons who received clinically verified AD diagnosis in between 2008 and 2011 in Finland and 1:1 age, sex and hospital district- matched comparison cohort were included. Hospitalizations and specialized healthcare visits were categorized by the main diagnosis, according to the ICD-10 chapters. AD and dementia were separated to their own category. The number of persons with visits and stays was calculated for every 6 months, irrespective of the frequency of visits/stays individual had during that time window. Furthermore, the relative distribution of the diagnosis categories was computed.ResultsAD cohort was more likely to have visits and stays during the 10-year period (OR 1.19, 95% CI 1.17–1.21). The number of persons with visits and stays peaked in AD cohort from 1.5 years before the diagnosis when the differences in relative distribution of different diagnosis categories also became evident. The largest differences were observed for visits/stays with cognitive disorders, symptoms of unspecified diseases and psychiatric disorders diagnoses, and those with missing diagnosis codes in the last time window before AD diagnosis.Conclusions and implicationsIncreased healthcare service use before AD diagnosis does not seem to arise from differences in specific diagnosis categories of ICD-10 such as diseases of the circulatory system, but from the higher frequency of visits and stays among persons with AD across diagnosis categories. Based on the relative distribution of diagnosis categories, the steep increase in healthcare service use just before and during the diagnostic process is likely due to prodromal symptoms and visits related to cognition.

Preventive DC-side decoupling: a control and operation concept to limit the impact of DC faults in offshore multi-terminal HVDC systems

For the integration of large amounts of offshore wind power, the interconnection of HVDC links towards multi-terminal networks is highly desired. However, its technical realisation remains a major challenge - especially in a multi-vendor environment. In addition to the challenges on the DC-side, it is crucial to guarantee that DC faults do not lead to an intolerable loss of power infeed to the connected AC grid(s). To restrict this power loss, a control and operation concept is presented in this paper which utilises state-of-the-art control and switchgear only: When the total wind power infeed exceeds the primary control reserve of the adjacent AC grid, the DC-side interconnection can be preventively decoupled via DC-side load commutation switches. At all other times, the multi-terminal network is operated in a coupled manner. Using (VDC/P)-droop controls and a DC grid controller, a decoupling sequence is implemented and validated via EMT simulations. The resulting DC fault behaviour is discussed along with technical requirements and limitations. The proposed concept could not only save costs, but also de-risk and accelerate the development of first offshore multi-terminal HVDC networks considering missing DC grid codes and a lack of coherent functional requirements for HVDC protection.

A 16-bit 300-kS/s foreground calibration SAR ADC with single-ended/differential configurable input modes

This paper presents a 16-bit 300-kS/s foreground calibration successive approximation register analog-to-digital converter (SAR ADC) with single-ended/differential configurable input modes. A foreground calibration method is implemented to correct mismatch errors in a capacitive digital-to-analog converter (CDAC) and reduce the capacitor size, thereby lowering the current consumption. The total capacitor size is reduced to 10 pF from the 485.6 pF required for 16-bit matching. A self-calibrated low-offset comparator is proposed to avoid the calibration being out of range. An improved merged capacitor switching technique is proposed to realize the configuration of the single-ended/differential input modes. Compared to active single-ended-to-differential converter solutions or single-ended SAR ADCs, the proposed solution achieves a smaller chip area and better power efficiency. A prototype was fabricated in 110-nm CMOS technology. It achieves a signal-to-noise and distortion ratio of 79.8 dB and integral nonlinearity of −1.5/+1.1 LSB with no missing codes. The power consumption of the ADC from a 3-V supply is 1.07 mW.

Design of a CMOS memristor emulator-based, self-adaptive spiking analog-to-digital data conversion as the lowest level of a self-x hierarchy

Abstract. The number of sensors used in modern devices is rapidly increasing, and the interaction with sensors demands analog-to-digital data conversion (ADC). A conventional ADC in leading-edge technologies faces many issues due to signal swings, manufacturing deviations, noise, etc. Designers of ADCs are moving to the time domain and digital designs techniques to deal with these issues. This work pursues a novel self-adaptive spiking neural ADC (SN-ADC) design with promising features, e.g., technology scaling issues, low-voltage operation, low power, and noise-robust conditioning. The SN-ADC uses spike time to carry the information. Therefore, it can be effectively translated to aggressive new technologies to implement reliable advanced sensory electronic systems. The SN-ADC supports self-x (self-calibration, self-optimization, and self-healing) and machine learning required for the internet of things (IoT) and Industry 4.0. We have designed the main part of SN-ADC, which is an adaptive spike-to-digital converter (ASDC). The ASDC is based on a self-adaptive complementary metal–oxide–semiconductor (CMOS) memristor. It mimics the functionality of biological synapses, long-term plasticity, and short-term plasticity. The key advantage of our design is the entirely local unsupervised adaptation scheme. The adaptation scheme consists of two hierarchical layers; the first layer is self-adapted, and the second layer is manually treated in this work. In our previous work, the adaptation process is based on 96 variables. Therefore, it requires considerable adaptation time to correct the synapses' weight. This paper proposes a novel self-adaptive scheme to reduce the number of variables to only four and has better adaptation capability with less delay time than our previous implementation. The maximum adaptation times of our previous work and this work are 15 h and 27 min vs. 1 min and 47.3 s. The current winner-take-all (WTA) circuits have issues, a high-cost design, and no identifying the close spikes. Therefore, a novel WTA circuit with memory is proposed. It used 352 transistors for 16 inputs and can process spikes with a minimum time difference of 3 ns. The ASDC has been tested under static and dynamic variations. The nominal values of the SN-ADC parameters' number of missing codes (NOMCs), integral non-linearity (INL), and differential non-linearity (DNL) are no missing code, 0.4 and 0.22 LSB, respectively, where LSB stands for the least significant bit. However, these values are degraded due to the dynamic and static deviation with maximum simulated change equal to 0.88 and 4 LSB and 6 codes for DNL, INL, and NOMC, respectively. The adaptation resets the SN-ADC parameters to the nominal values. The proposed ASDC is designed using X-FAB 0.35 µm CMOS technology and Cadence tools.

Successive approximation type resistance to digital converter

Successive approximation type direct resistance to digital converter (SAR RDC) topology is presented here. The conventional SAR type ADC is augmented with an additional opamp inverting summer and a standard resistance such that the digital output of the SAR directly indicates the value of an unknown resistance connected between the output of the digital to analog converter of the SAR RDC and the inverting input of the added inverting summer. Analysis of the proposed technique presented here helps select proper components at the design stage itself to obtain minimal errors in the measurement of the unknown resistor. The major advantage of the proposed scheme is that it retains the basic architecture of an SAR ADC and hence can be implemented with all available SAR ADC architectures, including the ability to install ‘auto-ranging’. An added advantage is that the value of the unknown resistor is determined in terms of a standard resistor and hence good accuracy is easily obtained. Simulation studies and results obtained from a prototype SAR RDC built and tested, establish the efficacy of the proffered technique. The results indicate a worst-case error of < ±0.2% and no missing codes.

Two-person game with hesitant fuzzy payoff: An application in MADM

Hesitant Fuzzy Set (HFS) permits the membership function having a collection of probable values which are more effective for modelling the real-life problems. Multiple Attribute Decision Making (MADM) process apparently assesses multiple conflicting attribute in decision making. In traditional decision making problems, each player is moving independently whereas in reality it is seen that each player aims to maximize personal profit which causes a negative impact on other player. MADM problem treats with candidate to the best alternative corresponding to the several attributes. Here, we present an MADM problem under hesitant fuzzy information and then transforming it into two-person matrix game, referred to herein as MADM game. The Technique for Order Preference by Similarity to Ideal Solution (TOPSIS) is one of the prominent approach for solving the MADM problems. In this work, we develop the TOPSIS based on Ordered Weighted Aggregation (OWA) operator and hybrid hesitant fuzzy normalized Euclidean distance.Please check whether short title on odd pages have been set correctly. Then the two approaches, namely Hybrid Hesitant Fuzzy Ordered Weighted Aggregation-TOPSIS (HHFOWA-TOPSIS) and the Linear Programming Problem (LPP) are applied to solve the formulated MADM game. For solving MADM game, we apply LPP by considering the various values of α,ψ, and HHFOWA-TOPSIS for finding the optimal alternative according to their scores.Please provide missing AMS classification codes. An investment selection problem is included to explain the feasibility and superiority of our formulated approaches. A comparison analysis is drawn among the obtained results which are derived from the two approaches. LPP and HHFOWA-TOPSIS provide the best alternative for the proposed problem. Finally, conclusions about our findings and outlooks are described.

From the Literature

From the Literature

Structuring clinical text with AI: Old versus new natural language processing techniques evaluated on eight common cardiovascular diseases

SummaryFree-text clinical notes in electronic health records are more difficult for data mining while the structured diagnostic codes can be missing or erroneous. To improve the quality of diagnostic codes, this work extracts diagnostic codes from free-text notes: five old and new word vectorization methods were used to vectorize Stanford progress notes and predict eight ICD-10 codes of common cardiovascular diseases with logistic regression. The models showed good performance, with TF-IDF as the best vectorization model showing the highest AUROC (0.9499–0.9915) and AUPRC (0.2956–0.8072). The models also showed transferability when tested on MIMIC-III data with AUROC from 0.7952 to 0.9790 and AUPRC from 0.2353 to 0.8084. Model interpretability was shown by the important words with clinical meanings matching each disease. This study shows the feasibility of accurately extracting structured diagnostic codes, imputing missing codes, and correcting erroneous codes from free-text clinical notes for information retrieval and downstream machine-learning applications.

Coverage of symptomatic toxicities from the common terminology criteria for adverse events (CTCAE) framework within the european organization for research and treatment of cancer (EORTC) item library.

e24117 Background: The EORTC Item Library is an online, interactive platform comprised of 950 distinct questions (items) from 67 different EORTC patient-reported outcome (PRO) questionnaires, covering a range of symptomatic toxicities, types of functioning, and impact on quality of life. These PROs provide a patient-centred perspective, complementing clinician adverse event (AE) reporting using classifications like the CTCAE. In order to summarize the coverage of symptomatic toxicities and facilitate the identification of items through use of a standardized framework, a mapping study was carried out, aimed at creating a coding system to classify EORTC items according to the CTCAE and link them to corresponding AEs. Methods: All items were searched for within the CTCAE (v5.0) using a deductive approach. Items were coded as linked if they were described within an AE’s title, description, or grading. Items not suitable for CTCAE coding were inductively assigned a descriptive classification. Descriptive classifications were also applied along with CTCAE codes when they provided additional information. Symptoms described in EORTC items but not located in the CTCAE were coded as missing and additional codes were assigned to highlight EORTC items capturing multiple underlying issues and diagnosis only CTCAE codes. Two raters independently coded 249 items and agreement was calculated. The remaining 701 items were coded by one rater and verified by the second, with a third introduced to discuss any discrepancies until a consensus was reached. Results: Agreement for raters following independent coding was 77.9% for at least one AE per item. In total, 625 (65.8%) items were linked to 208 different AEs. Fatigue was the most commonly linked AE, representing 4.9% of linked AEs. The majority of linked items were associated with one (65.6%) or two (23.5%) AEs, with some linked to three or more (10.9%). Multiple linkage resulted from different symptoms relating to the same issue/diagnosis or one symptom relating to multiple diagnoses. Two symptoms captured by six EORTC items but not found in the CTCAE were identified: bowel urgency and tenesmus. Seven descriptive non-CTCAE classifications emerged, with most of these covering the emotional impact of symptom, diagnosis, or treatment (33.6%) and information/satisfaction with care (31.7%). Nineteen items (2%) were linked to multiple underlying issues, and 43 (4.5%) to diagnosis only CTCAE codes. Conclusions: The EORTC Item Library provides extensive coverage of CTCAE symptomatic toxicities, along with other issues that are important to cancer patients, including emotional well-being and satisfaction with care services. Classifying symptomatic PRO items following the CTCAE clinical framework may facilitate future PRO selection and use in clinical trials and routine care.

Hepatitis B birth dose vaccination patterns in the Military Health System, 2014–2018

BackgroundSince 2005, the universal hepatitis B (HepB) birth dose has been recommended for all medically stable infants weighing ≥2,000 g at birth. The timing of the birth dose provides a critical safeguard and prevents infection among infants born to HBsAg-positive mothers not identified prenatally. We assess infant HepB vaccination in the U.S. Department of Defense’s Military Health System (MHS) to identify trends in vaccination coverage and sociodemographic factors associated with non-receipt of the birth dose, receiving the first HepB vaccine >3 days of life, and not receiving any HepB vaccine in the first 18 months of life utilizing parental refusal codes. To our knowledge, this is one of the first studies assessing trends in parental refusal of the HepB birth dose utilizing administrative claims parental refusal codes. MethodsWe conducted a retrospective cohort analysis of MHS live births from January 1, 2014 through December 31, 2018 utilizing administrative claims data. Data were included from 44 hospitals in 24 unique states, territories, or countries. We analyzed diagnosis codes for vaccine refusal and vaccination and current procedural terminology (CPT) codes to identify vaccination patterns. Generalized linear mixed effects models with a logit link were used to assess factors associated with vaccination patterns. ResultsHepB birth dose vaccination coverage increased from 79.6% in 2014 to 88.1% in 2018 (p < .0001). Refusal rates also increased from 3.7% in 2014 to 4.5% in 2018 (p < .0001). The percentage of patients with missing diagnosis codes for vaccine refusal or vaccination decreased from 16.7% in 2014 to 7.4% in 2018. Factors associated with non-receipt of the birth dose included earlier year of birth, white maternal race, higher maternal age, higher birth order, and longer infant length of stay in hospital. ConclusionVaccination coverage for HepB birth dose is high in the MHS and increased over time; concurrently, refusal rates also increased over time. Utilizing administrative claims data has the benefit of differentiating reasons for non-receipt of the birth dose over time.

The challenge of using routinely collected data to compare hospital admission rates by ethnic group: a demonstration project in Scotland.

Recording patients' ethnic group supports efforts to achieve equity in health care provision. Before the Equality Act (2010), recording ethnic group at hospital admission was poor in Scotland but has improved subsequently. We describe the first analysis of the utility of such data nationally for monitoring ethnic variation. We analysed all in-patient or day case hospital admissions in 2013. We imputed missing data using the most recent ethnic group recorded for a patient from 2009 to 2015. For episodes lacking an ethnic code, we attributed known ethnic codes proportionately. Using the 2011 Census population, we calculated rates and rate ratios for all-cause admissions and ischaemic heart diseases (IHDs) directly standardized for age. Imputation reduced missing ethnic group codes from 24 to 15% and proportionate redistribution to zero. While some rates for both all-cause and IHD admissions appeared plausible, unexpectedly low or high rates were observed for several ethnic groups particularly amongst White groups and newly coded groups. Completeness of ethnicity recoding on hospital admission records has improved markedly since 2010. However the validity of admission rates based on these data is variable across ethnic groups and further improvements are required to support monitoring of inequality.

The reliability of the ICD-AIS map in identifying serious road traffic injuries from the Helsinki Trauma Registry

ObjectiveThe EU has recommended that its member countries compile statistics on the number of serious road traffic injuries. In Finland, the number of seriously injured road traffic patients is assessed using the International Classification of Diseases, 10th Revision (ICD-10) and the automatic conversion tool (ICD-AIS map) developed by The Association for the Advancement of Automotive Medicine (AAAM). The aim of this study was to assess how reliably the ICD-AIS map identifies both serious injuries and seriously injured patients due to road traffic accidents. MethodsData was derived from the Helsinki Trauma Registry (HTR) and included 215 severe (New Injury Severity Score >15) trauma patients injured in road traffic accidents from the years 2016 and 2017. The severity ratings of injuries (Abbreviated Injury Scale, AIS 3+) and patients (Maximum Abbreviated Injury Scale, MAIS 3+) were determined by direct AIS coding of the HTR and were also generated by the ICD-AIS map based on ICD-10 injury codes. These two ratings were compared by injury mechanism and Injury Severity Score (ISS) body regions. The strength of agreement was described using Cohen’s κ. The most common injury codes with errors in severity rating by the ICD-AIS map were presented. ResultsThe number of seriously injured patients by the ICD-AIS map was 21% lower, and the number of serious injuries was 36% lower than the corresponding numbers by direct coding. The exact agreement of the injury ratings was 72% (κ = 0.44, 95% CI 0.42–0.46). Most of the conversion errors were due to the simplicity of the ICD-10 codes used in Finland compared to those used in the ICD-AIS map (ICD-10-CM) and the missing codes from the ICD-AIS map. The most frequent misclassifications were due to multiple rib fractures, visceral organ injuries, some open fractures of extremities, and specific head injuries. Missing codes were most common in face, chest, and limb injuries. ConclusionsThe ICD-10 injury codes presently used in Finland should be more specific to permit reliable conversion results by the ICD-AIS map. The problem with missing codes should be considered more closely. When implementing the ICD-11, all detailed injury codes should be introduced.

Feasibility of identifying important changes in care management resulting from cardiovascular magnetic resonance (CMR) using hospital episode data in patients who activate the primary percutaneous coronary intervention (PPCI) pathway

BackgroundWe determined whether it is feasible to identify important changes in care management resulting from cardiovascular magnetic resonance (CMR) in patients who activate the primary percutaneous coronary intervention (PPCI) pathway from hospital episode data, in order to construct a composite primary outcome (hypothesised to reduce the risk of major adverse cardiac-related events, MACE) to compare patients exposed to CMR or not.MethodsWe used Hospital Episode Statistics (HES) and Patient Episode Database for Wales (PEDW) to identify clinical events that reflected important changes in management in the year following the index admission in five subgroups of patients who activated the PPCI pathway recruited as part of a feasibility cohort study (n = 1655 with HES/PEDW data). For all subgroups, we identified frequency of events and time to the first event for each change in management.ResultsWe identified all clinical events (new diagnoses, additional diagnostic tests and procedures) except for medication prescriptions. Diagnostic tests were underestimated because most are carried out in outpatient clinics and outpatient datasets had missing procedure codes for 74% of patients (some tests done in hospital may also not be recorded). We successfully tabulated frequencies of events and distributions of times to first event for most changes in management by CMR status and in CMR / non CMR centres.ConclusionsIt is feasible to identify changes in care management between patients who have / do not have CMR within relevant patient subgroups. Further work to derive a weighting algorithm is required before attempting to combine the events in a composite endpoint.

Evaluation of Pedestrian/Bicycle Crash Injury Case Definitions for Use with NC DETECT

ObjectiveTo evaluate four ICD-10-CM based case definitions designed to capture pedestrian and bicycle crash-related emergency department (ED) visits in North Carolina’s statewide syndromic surveillance system, NC DETECT.IntroductionOver the last few decades, the United States has made considerable progress in decreasing the incidence of motor vehicle occupants injured and killed in traffic collisions.1 However, there is still a need for continued motor vehicle crash (MVC) injury surveillance, particularly for vulnerable road users, such as pedestrians and bicyclists. In NC, the average annual number of pedestrian-motor vehicle crashes increased by 13.5 percent during the period 2011-2015, as compared to 2006-2010.2 Therefore, the Carolina Center for Health Informatics (CCHI), as part of a NC Governor’s Highway Safety Program-funded project to improve statewide MVC injury surveillance, developed and evaluated four ICD-10-CM based case definitions for use with NC DETECT, NC’s statewide syndromic surveillance system.MethodsWe created four pedestrian/bicycle crash injury case definitions based on ICD-10-CM transportation codes (“V-codes”): Traffic-Related Pedestrian Crashes, Traffic/Non-Traffic-Related Pedestrian Crashes, Traffic-Related Bicycle Crashes, and Traffic/Non-Traffic-Related Bicycle Crashes. These definitions were based on the Centers for Disease Control and Prevention (CDC) “ICD-10-CM External Cause of Injury Codes”.3 We then applied these pedestrian/bicycle crash case definitions to 2016-2017 NC DETECT ED visit data and data obtained from a single NC Level I Trauma Center. Next, we linked the two data sources using the variables date of visit, time of visit, and medical record number. Since trauma center data are collected and verified by a designated trauma registrar, we considered the data obtained from the Level I Trauma Center to be the “gold standard”.ResultsThe linkage between the two data sources was successful, with 99.5% of all Level I Trauma Center records linking to ED visits in NC DETECT. However, we found discrepancies in the assignment of codes between the ED visit and Trauma Center data. For example, 47.5% of NC DETECT ED visits that linked to a pedestrian/bicycle crash record in the Trauma Center data, were missing an ICD-10-CM injury mechanism code of any category. Historically, the proportion of injury-related ED visits that were missing corresponding injury mechanism codes was low (&lt;15%). However, the transition from ICD-9-CM to ICD-10-CM increased the proportion of injury-related visits missing injury mechanism codes.4 Among the 92 NC DETECT ED visits missing injury mechanism codes, 35.9% contained a pedestrian/bicycle crash-related keyword in the Chief Complaint or Triage Note.Among the 100 linked records with valid ICD-10-CM injury mechanism codes, the percent agreement between the two data sources on whether the ED visit was a “pedestrian” or “bicycle” crash was 54.4% and 71.9%, respectively. Percent agreement decreased for “traffic” and “non-traffic” designations, however. The most common V-code assigned to misclassified pedestrian/bicycle crashes in the NC DETECT ED visit data was “V87.7XXA-Person injured in a collision between other specified motor vehicles (traffic)”.Although the linkage study used data obtained from only a single Level I Trauma Center and primarily a single facility in NC DETECT, we felt that the results of this limited linkage study were generalizable to statewide NC DETECT ED visit data. For example, many facilities in NC DETECT underreport injury mechanism codes. Therefore, we added pedestrian/bicycle crash injury-related keywords to the Traffic/Non-Traffic Pedestrian/Bicycle Crash Injury case definitions (Table 1). After inclusion of these keywords, the number of identified pedestrian and bicycle crash injury-related ED visits identified in NC DETECT increased by 16.9% and 57.9% from January-June 2018, respectively (Figure 1).ConclusionsPedestrian and bicycle crashes represent a major cause of MVC injury morbidity and mortality. Therefore, the development and evaluation of case definitions is key for the successful surveillance of these types of injuries. The inclusion of keywords can help account for some of the injury mechanism data missingness common to ED surveillance systems.