Establishing a valid cohort of patients with acromegaly by combining the National Patient Register with the Swedish Pituitary Register

PURPOSE

The aim of this study was to establish a valid national cohort of patients diagnosed with acromegaly by combining data from the general National Patient Register (NPR) and the disease-specific Swedish Pituitary Register (SPR).

METHODS

Patients ≥ 18 years of age at diagnosis of acromegaly reported from 1991 to 2018 who were registered in the NPR and/or SPR were included. The diagnosis of acromegaly was considered correct for patients identified in both registers or confirmed through chart review. Medical records were reviewed in two of Sweden´s six health care regions if the patient was reported only in the NPR. An algorithm for the NPR, with criteria requiring multiple diagnosis registrations and tumour and/or surgery codes, was constructed to reduce the number of patients to review in the remaining four regions.

RESULTS

A total of 1866 patients were identified. Among these, 938 were reported in both registers. After application of the algorithm and chart review, the diagnosis was confirmed for 83 of the 906 patients found only in the NPR. Among 22 patients only registered in the SPR, a review of medical records confirmed acromegaly in 13. This resulted in a total of 1034 cases with acromegaly during the study period. The incidence rate of acromegaly in Sweden 1991-2018 was calculated to 4.0/million/year in the entire population and 5.1/million/year among subjects ≥ 18 years of age.

CONCLUSION

The combination of the SPR and NPR established a valid cohort of patients diagnosed with acromegaly and increased the estimated incidence in Sweden.

A hybrid modeling framework for generalizable and interpretable predictions of ICU mortality across multiple hospitals

The development of reliable mortality risk stratification models is an active research area in computational healthcare. Mortality risk stratification provides a standard to assist physicians in evaluating a patient's condition or prognosis objectively. Particular interest lies in methods that are transparent to clinical interpretation and that retain predictive power once validated across diverse datasets they were not trained on. This study addresses the challenge of consolidating numerous ICD codes for predictive modeling of ICU mortality, employing a hybrid modeling approach that integrates mechanistic, clinical knowledge with mathematical and machine learning models . A tree-structured network connecting independent modules that carry clinical meaning is implemented for interpretability. Our training strategy utilizes graph-theoretic methods for data analysis, aiming to identify the functions of individual black-box modules within the tree-structured network by harnessing solutions from specific max-cut problems. The trained model is then validated on external datasets from different hospitals, demonstrating successful generalization capabilities, particularly in binary-feature datasets where label assessment involves extrapolation.

Validation of human immunodeficiency virus diagnosis codes among women enrollees of a U.S. health plan

BACKGROUND

Efficiently identifying patients with human immunodeficiency virus (HIV) using administrative health care data (e.g., claims) can facilitate research on their quality of care and health outcomes. No prior study has validated the use of only ICD-10-CM HIV diagnosis codes to identify patients with HIV.

METHODS

We validated HIV diagnosis codes among women enrolled in a large U.S. integrated health care system during 2010-2020. We examined HIV diagnosis code-based algorithms that varied by type, frequency, and timing of the codes in patients' claims data. We calculated the positive predictive values (PPVs) and 95% confidence intervals (CIs) of the algorithms using a medical record-confirmed diagnosis of HIV as the gold standard.

RESULTS

A total of 272 women with ≥ 1 HIV diagnosis code in the administrative claims data were identified and medical records were reviewed for all 272 women. The PPV of an algorithm classifying women as having HIV as of the first HIV diagnosis code during the observation period was 80.5% (95% CI: 75.4-84.8%), and it was 93.9% (95% CI: 90.0-96.3%) as of the second. Little additional increase in PPV was observed when a third code was required. The PPV of an algorithm based on ICD-10-CM-era codes was similar to one based on ICD-9-CM-era codes.

CONCLUSION

If the accuracy measure of greatest interest is PPV, our findings suggest that use of ≥ 2 HIV diagnosis codes to identify patients with HIV may perform well. However, health care coding practices may vary across settings, which may impact generalizability of our results.

To weight or not to weight? Studying the effect of selection bias in three large EHR-linked biobanks

Objective

To explore the role of selection bias adjustment by weighting electronic health record (EHR)-linked biobank data for commonly performed analyses.

Materials and methods

We mapped diagnosis (ICD code) data to standardized phecodes from three EHR-linked biobanks with varying recruitment strategies: All of Us (AOU; n=244,071), Michigan Genomics Initiative (MGI; n=81,243), and UK Biobank (UKB; n=401,167). Using 2019 National Health Interview Survey data, we constructed selection weights for AOU and MGI to be more representative of the US adult population. We used weights previously developed for UKB to represent the UKB-eligible population. We conducted four common descriptive and analytic tasks comparing unweighted and weighted results.

Results

For AOU and MGI, estimated phecode prevalences decreased after weighting (weighted-unweighted median phecode prevalence ratio [MPR]: 0.82 and 0.61), while UKB's estimates increased (MPR: 1.06). Weighting minimally impacted latent phenome dimensionality estimation. Comparing weighted versus unweighted PheWAS for colorectal cancer, the strongest associations remained unaltered and there was large overlap in significant hits. Weighting affected the estimated log-odds ratio for sex and colorectal cancer to align more closely with national registry-based estimates.

Discussion

Weighting had limited impact on dimensionality estimation and large-scale hypothesis testing but impacted prevalence and association estimation more. Results from untargeted association analyses should be followed by weighted analysis when effect size estimation is of interest for specific signals.

Conclusion

EHR-linked biobanks should report recruitment and selection mechanisms and provide selection weights with defined target populations. Researchers should consider their intended estimands, specify source and target populations, and weight EHR-linked biobank analyses accordingly.

Identifying patients with opioid use disorder using International Classification of Diseases (ICD) codes: Challenges and opportunities

BACKGROUND AND AIMS

International Classification of Diseases (ICD) diagnosis codes are often used in research to identify patients with opioid use disorder (OUD), but their accuracy for this purpose is not fully evaluated. This study describes application of ICD-10 diagnosis codes for opioid use, dependence and abuse from an electronic health record (EHR) data extraction using data from the clinics' OUD patient registries and clinician/staff EHR entries.

DESIGN

Cross-sectional observational study.

SETTING

Four rural primary care clinics in Washington and Idaho, USA.

PARTICIPANTS

307 patients.

MEASUREMENTS

This study used three data sources from each clinic: (1) a limited dataset extracted from the EHR, (2) a clinic-based registry of patients with OUD and (3) the clinician/staff interface of the EHR (e.g. progress notes, problem list). Data source one included records with six commonly applied ICD-10 codes for opioid use, dependence and abuse: F11.10 (opioid abuse, uncomplicated), F11.20 (opioid dependence, uncomplicated), F11.21 (opioid dependence, in remission), F11.23 (opioid dependence with withdrawal), F11.90 (opioid use, unspecified, uncomplicated) and F11.99 (opioid use, unspecified with unspecified opioid-induced disorder). Care coordinators used data sources two and three to categorize each patient identified in data source one: (1) confirmed OUD diagnosis, (2) may have OUD but no confirmed OUD diagnosis, (3) chronic pain with no evidence of OUD and (4) no evidence for OUD or chronic pain.

FINDINGS

F11.10, F11.21 and F11.99 were applied most frequently to patients who had clinical diagnoses of OUD (64%, 89% and 79%, respectively). F11.20, F11.23 and F11.90 were applied to patients who had a diagnostic mix of OUD and chronic pain without OUD. The four clinics applied codes inconsistently.

CONCLUSIONS

Lack of uniform application of ICD diagnosis codes make it challenging to use diagnosis code data from EHR to identify a research population of persons with opioid use disorder.

Accuracy of International Classification of Disease Coding Methods to Estimate Sepsis Epidemiology: A Scoping Review

PURPOSE

To provide an overview of various sepsis International Classification of Diseases (ICD) coding methods and their diagnostic accuracy.

METHODS

We undertook a systematic scoping review between 1991 and 2020 (search terms: sepsis, coding, and epidemiology) to include studies reporting the accuracy of a sepsis ICD coding method. Studies were grouped by ICD coding method, number of diagnostic accuracy parameters, ICD version, reference standard, design, country, setting, type of dataset and sepsis definition. ICD coding methods were categorised as explicit or implicit, with the explicit methods further divided into wide and narrow groups. Descriptive statistics were used to present data.

RESULTS

We analysed 17 studies, of which 16 (94.1%) used retrospective medical chart review as the reference standard for clinical sepsis, and eight (47.1%) used hospital administrative data to identify sepsis. There were 53 assessments of various ICD coding methods, with 32 (60.4%) of them being explicit and 21 (39.6%) implicit methods. The coding methods had a median sensitivity of <75% but a median specificity of >85%. However, a wide variation was noted in the diagnostic accuracy parameters of all ICD coding methods. Most of the studies showed high methodological quality.

CONCLUSION

None of the current ICD coding methods is optimal for identifying sepsis.

The applications of Australian-coded ICD-10 and ICD-10-AM data in research: A scoping review of the literature

Background: Australia uses the International Classification of Diseases (ICD-10) for mortality coding and its Australian Modification, ICD-10-AM, for morbidity coding. The ICD underpins surveillance (population health, mortality), health planning and research (clinical, epidemiological and others). ICD-10-AM also supports activity-based funding, thereby propelling realignment of the foci of clinical coding and, potentially, coded data's research utility. Objective: To conduct a scoping review of the literature exploring the use of ICD-10 and ICD-10-AM Australian-coded data in research. Research questions addressed herein: (1) What were the applications of ICD-10(-AM) Australian-coded data in published peer-reviewed research, 2012-2022? (2) What were the purposes of ICD-10(-AM) coded data within this context, as classified per a taxonomy of data use framework? Method: Following systematic Medline, Scopus and Cumulative Index to Nursing and Allied Health Literature database searches, a scoping literature review was conducted using PRISMA Extension for Scoping Reviews guidelines. References of a random 5% sample of within-scope articles were searched manually. Results were summarised using descriptive analyses. Results: Multi-stage screening of 2103 imported articles produced 636, including 25 from the references, for extraction and analysis; 54% were published 2019-2022; 50% within the largest five categories were published post-2019; 22% fell within the "Mental health and behavioural" category; 60.3% relied upon an ICD-10 modification. Articles were grouped by: research foci; relevant ICD chapter; themes per the taxonomy; purposes of the coded data. Observational study designs predominated: descriptive (50.6%) and cohort (34.6%). Conclusion: Researchers' use of coded data is extensive, robust and growing. Increasing demand is foreshadowed for ICD-10(-AM) coded data, and HIM-Coders' and Clinical Coders' expert advice to medical researchers.

Sepsis-coded hospitalisations and associated costs in Australia: a retrospective analysis

OBJECTIVE

To report trends in Australian hospitalisations coded for sepsis and their associated costs.

DESIGN

Retrospective analysis of Australian national hospitalisation data from 2002 to 2021.

METHODS

Sepsis-coded hospitalisations were identified using the Global Burden of Disease study sepsis-specific ICD-10 codes modified for Australia. Costs were calculated using Australian-Refined Diagnosis Related Group codes and National Hospital Cost Data Collection.

RESULTS

Sepsis-coded hospitalisations increased from 36,628 in 2002-03 to 131,826 in 2020-21, an annual rate of 7.8%. Principal admission diagnosis codes contributed 13,843 (37.8%) in 2002-03 and 44,186 (33.5%) in 2020-21; secondary diagnosis codes contributed 22,785 (62.2%) in 2002-03 and 87,640 (66.5%) in 2020-21. Unspecified sepsis was the most common sepsis code, increasing from 15,178 hospitalisations in 2002-03 to 68,910 in 2020-21. The population-based incidence of sepsis-coded hospitalisations increased from 18.6 to 10,000 population (2002-03) to 51.3 per 10,000 (2021-21); representing an increase from 55.1 to 10,000 hospitalisations in 2002-03 to 111.4 in 2020-21. Sepsis-coded hospitalisations occurred more commonly in the elderly; those aged 65 years or above accounting for 20,573 (55.6%) sepsis-coded hospitalisations in 2002-03 and 86,135 (65.3%) in 2020-21. The cost of sepsis-coded hospitalisations increased at an annual rate of 20.6%, from AUD199M (€127 M) in financial year 2012 to AUD711M (€455 M) in 2019.

CONCLUSION

Hospitalisations coded for sepsis and associated costs increased significantly from 2002 to 2021 and from 2012 to 2019, respectively.

Diagnostic accuracy, clinical characteristics, and prognostic differences of patients with acute myocarditis according to inclusion criteria

INTRODUCTION

The diagnosis of acute myocarditis (AM) is complex due to its heterogeneity and typically is defined by either Electronic Healthcare Records (EHRs) or advanced imaging and endomyocardial biopsy, but there is no consensus. We aimed to investigate the diagnostic accuracy of these approaches for AM.

METHODS

Data on ICD 10th Revision(ICD-10) codes corresponding to AM were collected from two hospitals and compared to CMR-confirmed or clinically suspected(CS) AM cases with respect to diagnostic accuracy, clinical characteristics, and all-cause mortality. Next, we performed a review of published AM studies according to inclusion criteria.

RESULTS

We identified 291 unique admissions with ICD-10 codes corresponding to AM in the first three diagnostic positions. The positive predictive value(PPV) of ICD-10 codes for CMR-confirmed or CS-AM was 36%, and patients with CMR-confirmed or CS AM had a lower all-cause mortality than those with a refuted diagnosis (P = 0.019). Using an unstructured approach, patients with CMR-confirmed and CS AM had similar demographics, comorbidity profiles and survival over a median follow-up of 52 months (P = 0.72). Our review of the literature confirmed our findings. Outcomes for patients included in studies using CMR-confirmed criteria were favourable compared to studies with EMB-confirmed AM cases.

CONCLUSION

ICD-10 codes have poor accuracy in identification of AM cases and should be used with caution in clinical research. There are important differences in management and outcomes of patients according to the selection criteria used to diagnose AM. Potential selection biases must be considered when interpreting AM cohorts and requires standardisation of inclusion criteria for AM studies.

pyPheWAS Explorer: a visualization tool for exploratory analysis of phenome-disease associations

Objective

To enable interactive visualization of phenome-wide association studies (PheWAS) on electronic health records (EHR).

Materials and Methods

Current PheWAS technologies require familiarity with command-line interfaces and lack end-to-end data visualizations. pyPheWAS Explorer allows users to examine group variables, test assumptions, design PheWAS models, and evaluate results in a streamlined graphical interface.

Results

A cohort of attention deficit hyperactivity disorder (ADHD) subjects and matched non-ADHD controls is examined. pyPheWAS Explorer is used to build a PheWAS model including sex and deprivation index as covariates, and the Explorer's result visualization for this model reveals known ADHD comorbidities.

Discussion

pyPheWAS Explorer may be used to rapidly investigate potentially novel EHR associations. Broader applications include deployment for clinical experts and preliminary exploration tools for institutional EHR repositories.

Conclusion

pyPheWAS Explorer provides a seamless graphical interface for designing, executing, and analyzing PheWAS experiments, emphasizing exploratory analysis of regression types and covariate selection.

Cannabis positivity rates in 17 emergency departments across the United States with varying degrees of marijuana legalization

BACKGROUND

Many states in the United States have progressed towards legalization of marijuana including decriminalization, medicinal and/or recreational use. We studied the impact of legalization on cannabis-related emergency department visits in states with varying degrees of legalization.

METHODS

Seventeen healthcare institutions in fifteen states (California, Colorado, Connecticut, Florida, Iowa, Kentucky, Maryland, Massachusetts, Missouri, New Hampshire, Oregon, South Carolina, Tennessee, Texas, Washington) participated. Cannabinoid immunoassay results and cannabis-related International Classification of Diseases (ninth and tenth versions) codes were obtained for emergency department visits over a 3- to 8-year period during various stages of legalization: no state laws, decriminalized, medical approval before dispensaries, medical dispensaries available, recreational approval before dispensaries and recreational dispensaries available. Trends and monthly rates of cannabinoid immunoassay and cannabis-related International Classification of Diseases code positivity were determined during these legalization periods.

RESULTS

For most states, there was a significant increase in both cannabinoid immunoassay and International Classification of Diseases code positivity as legalization progressed; however, positivity rates differed. The availability of dispensaries may impact positivity in states with medical and/or recreational approval. In most states with no laws, there was a significant but smaller increase in cannabinoid immunoassay positivity rates.

CONCLUSIONS

States may experience an increase in cannabis-related emergency department visits with progression toward marijuana legalization. The differences between states, including those in which no impact was seen, are likely multifactorial and include cultural norms, attitudes of local law enforcement, differing patient populations, legalization in surrounding states, availability of dispensaries, various ordering protocols in the emergency department, and the prevalence of non-regulated cannabis products.

Developing and validating natural language processing algorithms for radiology reports compared to ICD-10 codes for identifying venous thromboembolism in hospitalized medical patients

BACKGROUND

Identifying venous thromboembolism (VTE) from large clinical and administrative databases is important for research and quality improvement.

OBJECTIVE

To develop and validate natural language processing (NLP) algorithms to identify VTE from radiology reports among general internal medicine (GIM) inpatients.

METHODS

This cross-sectional study included GIM hospitalizations between April 1, 2010 and March 31, 2017 at 5 hospitals in Toronto, Ontario, Canada. We developed NLP algorithms to identify pulmonary embolism (PE) and deep venous thrombosis (DVT) from radiologist reports of thoracic computed tomography (CT), extremity compression ultrasound (US), and nuclear ventilation-perfusion (VQ) scans in a training dataset of 1551 hospitalizations. We compared the accuracy of our NLP algorithms, the previously-published "simpleNLP" tool, and administrative discharge diagnosis codes (ICD-10-CA) for PE and DVT to the "gold standard" manual review in a separate random sample of 4000 GIM hospitalizations.

RESULTS

Our NLP algorithms were highly accurate for identifying DVT from US, with sensitivity 0.94, positive predictive value (PPV) 0.90, and Area Under the Receiver-Operating-Characteristic Curve (AUC) 0.96; and in identifying PE from CT, with sensitivity 0.91, PPV 0.89, and AUC 0.96. Administrative diagnosis codes and the simple NLP tool were less accurate for DVT (ICD-10-CA sensitivity 0.63, PPV 0.43, AUC 0.81; simpleNLP sensitivity 0.41, PPV 0.36, AUC 0.66) and PE (ICD-10-CA sensitivity 0.83, PPV 0.70, AUC 0.91; simpleNLP sensitivity 0.89, PPV 0.62, AUC 0.92).

CONCLUSIONS

Administrative diagnosis codes are unreliable in identifying VTE in hospitalized patients. We developed highly accurate NLP algorithms to identify VTE from radiology reports in a multicentre sample and have made the algorithms freely available to the academic community with a user-friendly tool (https://lks-chart.github.io/CHARTextract-docs/08-downloads/rulesets.html#venous-thromboembolism-vte-rulesets).

Cannabis use assessment and documentation in healthcare: Priorities for closing the gap

Several factors, including the lack of a systematic cannabis use assessment within healthcare systems, have led to significant under-documentation of cannabis use and its correlates in medical records, the unpreparedness of clinicians, and poor quality of cannabis-related electronic health record data, limiting its utilization in research. Multiple steps are required to overcome the existing knowledge gaps and accommodate the health needs implied by the increasing cannabis use prevalence. These steps include (1) enhancing clinician and patient education on the importance of cannabis use assessment and documentation, (2) implementing a standardized approach for comprehensive cannabis use assessment within and across healthcare systems, (3) improving documentation of cannabis use and its correlates in medical records and electronic health records by building in prompts, (4) developing and validating reliable computable phenotypes of cannabis use, (5) conducting research utilizing electronic health data to study a wide array of related health outcomes, (6) and establishing evidence-based guidelines to inform clinical practices and policies. Integrating comprehensive cannabis use assessment and documentation within healthcare systems is necessary to enhance patient care and improve the quality of electronic health databases. Employing electronic health record data in cannabis-related research is crucial to accelerate research in light of the existing knowledge gaps on a wide array of health outcomes. Thus, improving and modernizing cannabis use assessment and documentation in healthcare is an integral step on which research conduct and evidence generation primarily rely.

Diagnoses during the First Three Years of Life for Children with Prenatal Opioid Exposure and Neonatal Abstinence Syndrome Using a Large Maternal Infant Data Hub

PURPOSE

The purpose of this study was to examine patterns of health and developmental outcomes in children with prenatal opioid exposure (POE) and neonatal abstinence syndrome (NAS) compared to children without exposure during the first three years of life.

DESIGN AND METHODS

This was a secondary data analysis of the Maternal and Infant Data Hub (MIDH), a de-identified dataset originating from the Midwest region of the United States, consisting of newborn billing records and corresponding maternal and child electronic medical records. For these analyses, the repository included data on more than 20,000 children born between 2013 and 2019. Diagnoses were identified with International Classification of Diseases, ninth and tenth Revision, Clinical Modification codes (ICD-9/10-CM). Firth logistic regression was used to assess whether incidence of each diagnosis code differed by exposure group.

RESULTS

Among 20,389 children in the dataset, 13,173 were unexposed; 455 were POE, and 199 were POE + NAS. There were significant differences in frequency of diagnoses between groups, specifically regarding growth and development, infection, mental health, musculoskeletal, neonatal, sensory, and social issues. When comparing exposed groups, children with POE + NAS experienced more negative health outcomes than children with only POE across all years.

CONCLUSIONS

This study implicates POE as a significant variable associated with many health and developmental outcomes of children during the first three years of life.

PRACTICE IMPLICATIONS

It is crucial to understand and identify health risks observed more frequently in exposed children during such a critical period of growth and brain development.

Comparative assessment of manual chart review and ICD claims data in evaluating immunotherapy-related adverse events

BACKGROUND

The aim of this retrospective study was to demonstrate that irAEs, specifically gastrointestinal and pulmonary, examined through International Classification of Disease (ICD) data leads to underrepresentation of true irAEs and overrepresentation of false irAEs, thereby concluding that ICD claims data are a poor approach to electronic health record (EHR) data mining for irAEs in immunotherapy clinical research.

METHODS

This retrospective analysis was conducted in 1,063 cancer patients who received ICIs between 2011 and 2017. We identified irAEs by manual review of medical records to determine the incidence of each of our endpoints, namely colitis, hepatitis, pneumonitis, other irAE, or no irAE. We then performed a secondary analysis utilizing ICD claims data alone using a broad range of symptom and disease-specific ICD codes representative of irAEs.

RESULTS

16% (n = 174/1,063) of the total study population was initially found to have either pneumonitis 3% (n = 37), colitis 7% (n = 81) or hepatitis 5% (n = 56) on manual review. Of these patients, 46% (n = 80/174) did not have ICD code evidence in the EHR reflecting their irAE. Of the total patients not found to have any irAEs during manual review, 61% (n = 459/748) of patients had ICD codes suggestive of possible irAE, yet were not identified as having an irAE during manual review.

DISCUSSION

Examining gastrointestinal and pulmonary irAEs through the International Classification of Disease (ICD) data leads to underrepresentation of true irAEs and overrepresentation of false irAEs.

Characterizing trends in methamphetamine-related health care use when there is no ICD code for "methamphetamine use disorder"

BACKGROUND AND AIMS

The recent surge in methamphetamine use highlights the need for timely data on its health effects and healthcare service use impact. However, there is no ICD code for methamphetamine use. This study quantifies the positive predictive value of ICD-9-CM and ICD-10-CM psychostimulant codes for methamphetamine use.

METHODS

A retrospective chart review of 220 adults aged 18 and older who had an inpatient admission with a psychostimulant-associated billing diagnosis at an urban safety-net hospital. Diagnoses were categorized as either methamphetamine-related or involving another specific psychostimulant. The positive predictive value of both ICD-9-CM or ICD-10-CM psychostimulant diagnosis codes for methamphetamine use was calculated.

RESULTS

ICD-9-CM and ICD-10-CM psychostimulant codes had high positive predictive values of 78.2% (95% CI 70.3%-86.0%) and 85.5% (95% CI 78.8%-92.1%), respectively, for methamphetamine use. The most common non-methamphetamine psychostimulant in our cohort was khat, a cathinone-containing plant native to East Africa, accounting for psychostimulant-related diagnosis in 16 of the 220 hospitalizations.

CONCLUSIONS

The high predictive values of psychostimulant codes for methamphetamine use support the application of administrative data in measuring methamphetamine-related healthcare use, as well as co-morbid health conditions and treatment patterns.

Diagnostic Accuracy of Health Care Administrative Diagnosis Codes to Identify Nontuberculous Mycobacteria Disease: A Systematic Review

Background

Health care administrative database research frequently uses standard medical codes to identify diagnoses or procedures. The aim of this review was to establish the diagnostic accuracy of codes used in administrative data research to identify nontuberculous mycobacterial (NTM) disease, including lung disease (NTMLD).

Methods

We searched Ovid Medline, Embase, Scopus, Cochrane Database of Systematic Reviews, Cochrane Central Register of Controlled Trials, and ClinicalTrials.gov from inception to April 2019. We included studies assessing the diagnostic accuracy of International Classification of Diseases, 9th edition, Clinical Modification (ICD-9-CM) diagnosis codes to identify NTM disease and NTMLD. Studies were independently assessed by 2 researchers, and the Quality Assessment of Diagnostic Accuracy Studies 2 tool was used to assess bias and quality.

Results

We identified 5549 unique citations. Of the 96 full-text articles reviewed, 7 eligible studies of moderate quality (3730 participants) were included in our review. The diagnostic accuracy of ICD-9-CM diagnosis codes to identify NTM disease varied widely across studies, with positive predictive values ranging from 38.2% to 100% and sensitivity ranging from 21% to 93%. For NTMLD, 4 studies reported diagnostic accuracy, with positive predictive values ranging from 57% to 64.6% and sensitivity ranging from 21% to 26.9%.

Conclusions

Diagnostic accuracy measures of codes used in health care administrative data to identify patients with NTM varied across studies. Overall the positive predictive value of ICD-9-CM diagnosis codes alone is good, but the sensitivity is low; this method is likely to underestimate case numbers, reflecting the current limitations of coding systems to capture NTM diagnoses.

Identifying individuals with opioid use disorder: Validity of International Classification of Diseases diagnostic codes for opioid use, dependence and abuse

BACKGROUND

Policy evaluations and health system interventions often utilize International Classification of Diseases (ICD) codes of opioid use, dependence, and abuse to identify individuals with opioid use disorder (OUD) and assess receipt of evidence-based treatments. However, ICD codes may not map directly onto the Diagnostic and Statistical Manual of Mental Disorder (DSM-5) OUD criteria. This study investigates the positive predictive value of ICD codes in identifying patients with OUD.

METHODS

We conducted a clinical chart review on a national sample of 520 Veterans assigned ICD-9 or ICD-10 codes for opioid use, dependence, or abuse from 2012 to 2017. We extracted evidence of DSM-5 OUD criteria and opioid misuse from clinical documentation in the month preceding and three months following initial ICD code listing, and categorized patients into: 1) high likelihood of OUD, 2) limited aberrant opioid use, 3) prescribed opioid use without evidence of aberrant use, and 4) insufficient information. Positive predictive value was calculated as the percentage of individuals with these ICD codes meeting high likelihood of OUD criteria upon chart review.

RESULTS

Only 57.7 % of patients were categorized as high likelihood of OUD; 16.5 % were categorized as limited aberrant opioid use, 18.9 % prescribed opioid use without evidence of aberrant use, and 6.9 % insufficient information.

CONCLUSIONS

Patients assigned ICD codes for opioid use, dependence, or abuse often lack documentation of meeting OUD criteria. Many receive long-term opioid therapy for chronic pain without evidence of misuse. Robust methods of identifying individuals with OUD are crucial to improving access to clinically appropriate treatment.

Automated screening of natural language in electronic health records for the diagnosis septic shock is feasible and outperforms an approach based on explicit administrative codes

PURPOSE

Identification of patients for epidemiologic research through administrative coding has important limitations. We investigated the feasibility of a search based on natural language processing (NLP) on the text sections of electronic health records for identification of patients with septic shock.

MATERIALS AND METHODS

Results of an explicit search strategy (using explicit concept retrieval) and a combined search strategy (using both explicit and implicit concept retrieval) were compared to hospital ICD-9 based administrative coding and to our department's own prospectively compiled infection database.

RESULTS

Of 8911 patients admitted to the medical or surgical ICU, 1023 (11.5%) suffered from septic shock according to the combined search strategy. This was significantly more than those identified by the explicit strategy (518, 5.8%), by hospital administrative coding (549, 5.8%) or by our own prospectively compiled database (609, 6.8%) (p < .001). Sensitivity and specificity of the automated combined search strategy were 72.7% (95%CI 69.0%-76.2%) and 93.0% (95%CI 92.4%-93.6%), compared to 56.0% (95%CI 52.0%-60.0%) and 97.5% (95%CI 97.1%-97.8%) for hospital administrative coding.

CONCLUSIONS

An automated search strategy based on a combination of explicit and implicit concept retrieval is feasible to screen electronic health records for septic shock and outperforms an administrative coding based explicit approach.

Coding reliability and agreement of International Classification of Disease, 10th revision (ICD-10) codes in emergency department data

Introduction

Administrative health data from emergency departments play important roles in understanding health needs of the public and reasons for health care resource use. International Classification of Disease (ICD) diagnostic codes have been widely used to code reasons of clinical encounters for administrative purposes in emergency departments.

Objective

The purpose of the study is to examine the coding agreement and reliability of ICD diagnosis codes in emergency department records through auditing the routinely collected data.

Methods

We randomly sampled 1 percent of records (n=1636) between October and December 2013 from 11 emergency departments in Alberta, Canada. Auditors were employed to review the same chart and independently assign main diagnosis codes. We assessed coding agreement and reliability through comparison of codes assigned by auditors and hospital coders using proportion of agreement and Cohen's kappa. Error analysis was conducted to review diagnosis codes with disagreement and categorized them into six groups.

Results

Overall, the agreement was 86.5% and 82.2% at 3 and 4 digits levels respectively, and reliability was 0.86 and 0.82 respectively. Variations of agreement and reliability were identified across different emergency departments. The major two categories of coding discrepancy were the use of different codes for same condition (23.6%) and the use of codes at different levels of specificity (20.9%).

Conclusions

Diagnosis codes in emergency departments show high agreement and reliability, although there are variations of coding quality across different hospitals. Stricter coding guidelines regarding the use of unspecified codes are needed to enhance coding consistency.

Analysis of Age Distribution and Disease Presentation of 1269 Patients with Sarcoidosis

Objective

While the incidence of sarcoidosis peaks between 20 and 39 years, it is comparatively low in elderly subjects. We sought to determine whether there are age-dependent differences in the demographic and laboratory characteristics of patients with sarcoidosis.

Materials and Methods

We retrospectively collected information from our database using the International Classification of Disease (ICD) diagnostic code D86 between 2008 and 2014. Patients were divided into three groups: 20-39 years old (Group 1), 40-59 years old (Group 2), and 60-80 years old (Group 3).

Results

A total of 3988 patients with code of D86 were included in the study. After the exclusion of non-eligible patients, the number of cases in Groups 1, 2, and 3 were 276, 641, and 352, respectively. The groups were compared according to demographic characteristics, ICD diagnostic codes, and laboratory parameters. The ratio of female patients was significantly higher in Group 3 than in Groups 1 and 2 (p=0.000). There was no difference in diagnostic codes of the ICD subgroups between groups (p=0.19). While the level of blood-urea nitrogen was significantly higher in Group 3 patients than in other groups (p=0.000), serum angiotensin-converting enzyme (ACE) values were found to be significantly low in Group 3 (p=0.010). The mean ACE values did not differ between females and males (50.8±39.3 and 59.1±45.5 mg/dL, respectively) (p=0.18).

Conclusion

The majority of patients with sarcoidosis were female in all age groups and pulmonary sarcoidosis was the most common presentation of the disease. Elderly patients (≥60 years) with sarcoidosis had lower serum ACE levels than younger patients.

Under-triage in telephone consultation is related to non-normative symptom description and interpersonal communication: a mixed methods study

Background

Telephone consultation and triage are used to limit the workload on emergency departments. Lack of visual cues and clinical tests put telephone consultations to a disadvantage compared to face-to-face consultations increasing the risk of under-triage. Under-triage occurs in telephone triage; however why under-triage happens is not explored yet. The aim of the study was to describe situations of under-triage in context, to assess the quality of under-triaged calls, and to identify communication patterns contributing to under-triage in a regional OOH service in the capital region of Denmark.

Methods

Explanatory simultaneous mixed method with thematic analysis and descriptive statistics was chosen. The study was carried out in an Out-Of-Hours service (OOH) in the Capital Region of Denmark, Copenhagen. Under-triage was defined as Potentially Under-Triaged Calls (PUTC) by specific criteria to an OOH Hotline, and identification by integration of three databases: Medical Hotline database, Emergency number database, including the Ambulance database, and electronic patient records. Distribution of PUTC were carried out using ICD-10 codes to identify diagnosis and main themes identified by qualitative analysis of audio recorded under-triaged calls. Study period was October 15^th to November 30^th 2014.

Results

Three hundred twenty seven PUTC were identified, representing 0.04% of all calls (n = 937.056) to the OOH. Distribution of PUTC according to diagnoses was: digestive (24%), circulatory (19%), respiratory (15%) and all others (42%). Thematic analysis of the voice logs suggested that inadequate communication and non-normative symptom description contributed to under-triage.

Discussion

The incidence of potentially under-triage is low (0.04%). However, the over-representation of digestive, circulatory, and respiratory diagnoses might suggest that under-triage is related to inadequate symptom description. We recommend that caller and call-handler collaborate systematically on problem identification and negotiate non-normative symptom description.

Conclusion

The incidence of under-triage is low (0.04%). However, the over-representation of digestive, circulatory, and respiratory diagnoses might suggest that under-triage is related to inadequate symptom description. We recommend that caller and call-handler collaborate systematically on problem identification and negotiate non-normative symptom description.

Using Electronic Medical Record to Identify Patients With Dyslipidemia in Primary Care Settings: International Classification of Disease Code Matters From One Region to a National Database

Objective:

To assess the validity of the International Classification of Disease (ICD) codes for identifying patients with dyslipidemia in electronic medical record (EMR) data

Methods:

The EMRs of patients receiving primary care in St. John’s, Newfoundland and Labrador (NL), Canada, were retrieved from the Canadian Primary Care Sentinel Surveillance Network database. International Classification of Disease codes were first compared with laboratory lipid data as an independent criterion standard, and next with a “comprehensive criterion standard,” defined as any existence of abnormal lipid test, lipid-lowering medication record, or dyslipidemia ICD codes. The ability of ICD coding alone or combined with other components was evaluated against the two criterion standards using receiver operating characteristic (ROC) analysis, sensitivity, specificity, negative predictive value (NPV) and Kappa agreement. (No specificity was reported for the comparison of ICD codes against the comprehensive criterion standard as this naturally leads to 100% specificity.)

Results:

The ICD codes led to a poor outcome when compared with the serum lipid levels (sensitivity, 27%; specificity, 76%; PPV, 71%; NPV, 33%; Kappa, 0.02; area under the receiver operating characteristic curve (AUC), 0.51) or with the comprehensive criterion standard (sensitivity, 32%; NPV, 25%; Kappa, 0.15; AUC, 66%). International Classification of Disease codes combined with lipid-lowering medication data also resulted in low sensitivity (51.2%), NPV (32%), Kappa (0.28), and AUC (75%). The addition of laboratory lipid levels to ICD coding marginally improved the algorithm (sensitivity, 94%; NPV, 79%; Kappa, 0.85; AUC, 97%).

Conclusions:

The use of ICD coding, either alone or in combination with laboratory data or lipid-lowering medication records, was not an accurate indicator in identifying dyslipidemia.

Increased consultation frequency in primary care, a risk marker for cancer: a case-control study

OBJECTIVE

To identify early diagnostic profiles such as diagnostic codes and consultation patterns of cancer patients in primary care one year prior to cancer diagnosis.

DESIGN

Total population-based case-control study.

SETTING AND SUBJECTS

4562 cancer patients and 17,979 controls matched by age, sex, and primary care unit. Data were collected from the Swedish Cancer Register and the Regional Healthcare Database.

METHOD

We identified cancer patients in the Västra Götaland Region of Sweden diagnosed in 2011 with prostate, breast, colorectal, lung, gynaecological, and skin cancers including malignant melanoma. We studied the symptoms and diagnoses identified by diagnostic codes during a diagnostic interval of 12 months before the cancer diagnosis.

MAIN OUTCOME MEASURES

Consultation frequency, symptom density by cancer type, prevalence and odds ratios (OR) for the diagnostic codes in the cancer population as a whole.

RESULTS

The diagnostic codes with the highest OR were unspecified lump in breast, neoplasm of uncertain behaviour, and abnormal serum enzyme levels. The codes with the highest prevalence were hyperplasia of prostate, other skin changes and abdominal and pelvic pain. The frequency of diagnostic codes and consultations in primary care rose in tandem 50 days before diagnosis for breast and gynaecological cancer, 60 days for malignant melanoma and skin cancer, 80 days for prostate cancer and 100 days for colorectal and lung cancer.

CONCLUSION

Eighty-seven percent of patients with the most common cancers consulted a general practitioner (GP) a year before their diagnosis. An increase in consultation frequency and presentation of any symptom should raise the GP's suspicion of cancer. Key points Knowledge about the prevalence of early symptoms and other clinical signs in cancer patients in primary care remains insufficient. • Eighty-seven percent of the patients with the seven most common cancers consulted a general practitioner 12 months prior to cancer diagnosis. • Both the frequency of consultation and the number of symptoms and diseases expressed in diagnostic codes rose in tandem 50-100 days before the cancer diagnosis. • Unless it is caused by a previously known disease, an increased consultation rate for any symptom should result in a swift investigation or referral from primary care to confirm or exclude cancer.

Kawasaki disease: an epidemiological study in central Italy

BACKGROUND

Kawasaki disease (KD) is a systemic vasculitis with an acute and self-limited course. The incidence of KD differs widely among ethnic groups and is higher in the Asian population. In Italy, no recent data are available. Our purpose is to define the epidemiology of Kawasaki disease in the years 2008-2013 in children aged < 14 years in the Italian regions of Tuscany and Emilia Romagna through administrative data.

METHODS

We studied the epidemiology of KD in the years 2008-2013 in children 0-14 years old resident in Tuscany and in Emilia Romagna regions using hospital ICD-9 discharge codes with a thorough data cleaning for duplicates.

RESULTS

The distribution of the KD patients across ages was similar for the two regions with a peak in the second year of life. When considering data of the two regions together, the rate of incidence was 17.6 for 100,000 children under 5 years. For both Regions the incidence rose slightly during the study period and had a seasonal distribution, with higher incidence in spring and winter.

CONCLUSION

This is the first Italian study performed through the use of administrative data. Figures are in line but slightly higher than those published in other European countries.

A method for reclassifying cause of death in cases categorized as "event of undetermined intent"

Background

We present a method for reclassifying external causes of death categorized as “event of undetermined intent” (EUIs) into non-transport accidents, suicides, or homicides. In nations like Russia and the UK the absolute number of EUIs is large, the EUI death rate is high, or EUIs comprise a non-trivial proportion of all deaths due to external causes. Overuse of this category may result in (1) substantially underestimating the mortality rate of deaths due to specific external causes and (2) threats to the validity of studies of the patterns and causes of external deaths and of evaluations of the impact of interventions meant to reduce them.

Methods

We employ available characteristics about the deceased and the event to estimate the most likely cause of death using multinomial logistic regression. We use the set of known non-transport accidents, suicides, and homicides to calculate an mlogit-based linear score and an estimated classification probability (ECP). This ECP is applied to EUIs, with varying levels of minimal classification probability. We also present an optional second step that employs a population-level adjustment to reclassify deaths that remain undetermined (the proportion of which varies based on the minimal classification probability). We illustrate our method by applying it to Russia. Between 2000 and 2011, 521,000 Russian deaths (15 % percent of all deaths from external causes) were categorized as EUIs. We used data from anonymized micro-data on the ~3 million deaths from external causes. Our reclassification model used 10 decedent and event characteristics from the computerized death records.

Results

Results show that during this period about 14 % of non-transport accidents, 13 % of suicides, and 33 % of homicides were officially categorized as EUIs. Our findings also suggest that 2011 levels of non-transport accidents and suicides would have been about 24 % higher and of homicide about 82 % higher than that reported by official vital statistics data.

Conclusions

Overuse of the external cause of death classification “event of undetermined intent” may indicate questionable quality of mortality data on external causes of death. This can have wide-ranging implications for families, medical professionals, the justice system, researchers, and policymakers. With our classification probability set as equal to or higher than 0.75, we were able to reclassify about two-thirds of EUI deaths in our sample. Our optional additional step allowed us to redistribute the remaining unclassified EUIs. Our method can be applied to data from any nation or sub-national population in which the EUI category is employed.

Prevalence of alcohol-related pathologies at autopsy: Estonian Forensic Study of Alcohol and Premature Death

AIMS

Alcohol can induce diverse serious pathologies, yet this complexity may be obscured when alcohol-related deaths are classified according to a single underlying cause. We sought to quantify this issue and its implications for analysing mortality data.

DESIGN, SETTING AND PARTICIPANTS

Cross-sectional study included 554 men aged 25-54 in Estonia undergoing forensic autopsy in 2008-09.

MEASUREMENTS

Potentially alcohol-related pathologies were identified following macroscopic and histological examination. Alcohol biomarkers levels were determined. For a subset (26%), drinking behaviour was provided by next-of-kin. The Estonian Statistics Office provided underlying cause of death.

FINDINGS

Most deaths (75%) showed evidence of potentially alcohol-related pathologies, and 32% had pathologies in two or more organs. The liver was most commonly affected [60.5%, 95% confidence interval (CI) = 56.3-64.6] followed by the lungs (18.6%, 95% CI = 15.4-22.1), stomach (17.5%, 95% CI = 14.4-20.9), pancreas (14.1%, 95% CI = 11.3-17.3), heart (4.9%, 95% CI = 3.2-7.0) and oesophagus (1.4%, 95% CI = 0.6-2.8). Only a minority with liver pathology had a second pathology. The number of pathologies correlated with alcohol biomarkers (phosphatidylethanol, gamma-glytamyl transpeptidase in blood, ethylglucuronide, ethylsulphate in urine). Despite the high prevalence of liver pathology, few deaths had alcoholic liver disease specified as the underlying cause.

CONCLUSION

The majority of 554 men aged 25-54 undergoing forensic autopsy in Estonia in 2008-09 showed evidence of alcohol-related pathology. However, the recording of deaths by underlying cause failed to capture the scale and nature of alcohol-induced pathologies found.

Validity of congenital malformation diagnostic codes recorded in Québec's administrative databases

PURPOSE

To assess the validity of the diagnostic codes of congenital malformations (CMs) recorded in two of Québec's administrative databases.

METHODS

A cohort of pregnancies and infants born to asthmatic and non-asthmatic women in 1990-2002 was reconstructed using Québec's administrative databases. From this cohort, we selected 269 infants with a CM and 144 without CM born to asthmatic women, together with 284 and 138 infants, respectively, born to non-asthmatic women. The diagnoses of CMs recorded in the databases were compared with the diagnoses written by the physicians in the infants' medical charts. The positive predictive values (PPV) and negative predictive values (NPV) for all, major, and several specific CMs were estimated.

RESULTS

The PPVs for all CMs and major CMs were 82.2% (95% confidence interval (CI): 78.5%-85.9%) and 78.1% (74.1%-82.1%), respectively, in the asthmatic group and were 79.2% (75.4%-83.1%) and 69.0% (64.6%-73.4%), respectively, in the non-asthmatic group. PPVs >80% were found for several specific CMs, including cardiac, cleft, and limb CMs in both groups. The NPV for any CM was 88.2% (95% CI: 85.1%-91.3%) in the asthmatic group and 94.2% (92.2%-96.2%) in the non-asthmatic group.

CONCLUSIONS

Québec's administrative databases are valid tools for epidemiological research of CMs. The results were similar between infants born to women with and without asthma.