Accuracy of diagnostic coding for Medicare patients under the prospective-payment system

Behaviours and drivers of diagnosis-related group upcoding in China: A mixed-methods study

As a highly destructive gaming behaviour in Diagnosis-Related Group (DRG), upcoding has garnered increasing scholarly attention. This study considers the prevalence, types and risk characteristics of upcoding during the pilot implementation of DRG payments in China, and it also explores the drivers of upcoding and provides corresponding policy recommendations for improving the system. Quantitative research data were sourced from the DRG payment audit database in City Z between the dates of June 1, 2019 and May 31, 2020, encompassing audit results comprising 200 medical records randomly selected from 28 hospitals. Qualitative research methods were used, including semi-structured interviews conducted with 10 stakeholders with interests in the DRG payment system, and thematic framework of the consequent data. 5,157 (92.01%) valid records were re-abstracted. 666 (12.91%) evaluated records were found to be upcoded, resulting in an additional payment at a rate of 45.27%. Several factors emerged as shedding light on the probability of upcoding, including cases with comorbidities, those undergoing non-operating room procedures and internal medical treatments, cases in for-profit hospitals and cases in tertiary hospitals. The main drivers of upcoding were found to be financial and administrative pressures, dysfunctional attitudes towards upcoding, technical facilitation and lack of supervision. This paper provides a comprehensive analysis of the behaviours and drivers of DRG upcoding in China, considering the unique hospital management system and incentive mechanisms in place. The results demonstrate that, following the initiation of the DRG payment system, providers have begun to engage in upcoding behaviour under various drivers, leading to additional health care expenditures and undermining the effectiveness of the scheme. In terms of mounting a response to this behaviour, understanding it and what drives it can aid in its prevention. This study suggests implementing intelligent audits to strengthen supervision and supporting hospitals in cost management.

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

BACKGROUND

The 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.

METHODS

All 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.

RESULTS

Data 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.

CONCLUSIONS

We 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.

Measuring misclassification and sample bias in passive surveillance systems: Improving prevalence estimates of critical congenital heart defects in state-based passive surveillance systems

OBJECTIVES

We assessed reporting misclassification for 12 critical congenital heart defects (CCHDs) identified through administrative diagnosis codes within a passive surveillance system. We measured the effect of misclassification on prevalence estimation. Lastly, we investigated a sample-based review strategy to estimate surveillance misclassification resulting from administrative diagnosis codes for case detection.

METHODS

We received 419 reports of CCHDs between 2007 and 2018; 414 were clinically reviewed. We calculated confirmation probabilities to assess misclassification and adjust prevalence estimates. Random samples of reported cases were taken at proportions between 20% and 90% for each condition to assess sample bias. Sampling was repeated 1000 times to measure sample-estimate variability.

RESULTS

Misclassification ranged from a low of 19% (n = 4/21) to a high of 84% (n = 21/25). Unconfirmed prevalence rates ranged between one and six cases per 10,000 live births, with some conditions significantly higher than national estimates. However, confirmed rates were either lower or comparable to national estimates.

CONCLUSION

Passive birth defect surveillance programs that rely on administrative diagnosis codes for case identification of CCHDs are subject to misclassification that bias prevalence estimates. We showed that a sample-based review could improve the prevalence estimates of 12 cardiovascular conditions relative to their unconfirmed prevalence rates.

A Data-Driven Approach to Defining Risk-Adjusted Coding Specificity Metrics for a Large U.S. Dementia Patient Cohort

Medical coding impacts patient care quality, payor reimbursement, and system reliability through the precision of patient information documentation. Inadequate coding specificity can have significant consequences at administrative and patient levels. Models to identify and/or enhance coding specificity practices are needed. Clinical records are not always available, complete, or homogeneous, and clinically driven metrics to assess medical practices are not logistically feasible at the population level, particularly in non-centralized healthcare delivery systems and/or for those who only have access to claims data. Data-driven approaches that incorporate all available information are needed to explore coding specificity practices. Using N = 487,775 hospitalization records of individuals diagnosed with dementia and discharged in 2022 from a large all-payor administrative claims dataset, we fitted logistic regression models using patient and facility characteristics to explain the coding specificity of principal and secondary diagnoses of dementia. A two-step approach was produced to allow for the flexible clustering of patient-level outcomes. Model outcomes were then used within a Poisson binomial model to identify facilities that over- or under-specify dementia diagnoses against healthcare industry standards across hospitalizations. The results indicate that multiple factors are significantly associated with dementia coding specificity, especially for principal diagnoses of dementia (AUC = 0.727). The practical use of this novel risk-adjusted metric is demonstrated for a sample of facilities and geospatially via a U.S. map. This study's findings provide healthcare facilities with a benchmark for assessing coding specificity practices and developing quality enhancements to align with healthcare industry standards, ultimately contributing to better patient care and healthcare system reliability.

Measuring accuracy of International Classification of Diseases codes in identification of patients with non-cystic fibrosis bronchiectasis

Introduction

Non-cystic fibrosis bronchiectasis is a disease which is increasing in incidence and prevalence worldwide. The incidence of the disease is frequently estimated using databases that rely on International Classification of Diseases, ninth and tenth revisions, clinical modification (ICD-9-CM/ICD-10-CM) discharge diagnoses. Code accuracy has proved to be a major issue for other diagnoses using ICD codes. This study aims to investigate the accuracy of the ICD codes for the diagnosis of non-cystic fibrosis bronchiectasis.

Methods

This is a retrospective diagnostic accuracy study which compares the radiologist's diagnosis of bronchiectasis with the ICD code reflection of that diagnosis at discharge.

Results

Sensitivities were 34% (same for both ICD-9-CM and ICD-10-CM windows) and specificities ranged from 69% for the ICD-9-CM window to 81% for ICD-10-CM window.

Conclusion

We observed that ICD codes are an insufficient method to identify patients with a radiologist diagnosis of bronchiectasis.

PetBERT: automated ICD-11 syndromic disease coding for outbreak detection in first opinion veterinary electronic health records

Effective public health surveillance requires consistent monitoring of disease signals such that researchers and decision-makers can react dynamically to changes in disease occurrence. However, whilst surveillance initiatives exist in production animal veterinary medicine, comparable frameworks for companion animals are lacking. First-opinion veterinary electronic health records (EHRs) have the potential to reveal disease signals and often represent the initial reporting of clinical syndromes in animals presenting for medical attention, highlighting their possible significance in early disease detection. Yet despite their availability, there are limitations surrounding their free text-based nature, inhibiting the ability for national-level mortality and morbidity statistics to occur. This paper presents PetBERT, a large language model trained on over 500 million words from 5.1 million EHRs across the UK. PetBERT-ICD is the additional training of PetBERT as a multi-label classifier for the automated coding of veterinary clinical EHRs with the International Classification of Disease 11 framework, achieving F1 scores exceeding 83% across 20 disease codings with minimal annotations. PetBERT-ICD effectively identifies disease outbreaks, outperforming current clinician-assigned point-of-care labelling strategies up to 3 weeks earlier. The potential for PetBERT-ICD to enhance disease surveillance in veterinary medicine represents a promising avenue for advancing animal health and improving public health outcomes.

Performance of a Computational Phenotyping Algorithm for Sarcoidosis Using Diagnostic Codes in Electronic Medical Records: Case Validation Study From 2 Veterans Affairs Medical Centers

BACKGROUND

Electronic medical records (EMRs) offer the promise of computationally identifying sarcoidosis cases. However, the accuracy of identifying these cases in the EMR is unknown.

OBJECTIVE

The aim of this study is to determine the statistical performance of using the International Classification of Diseases (ICD) diagnostic codes to identify patients with sarcoidosis in the EMR.

METHODS

We used the ICD diagnostic codes to identify sarcoidosis cases by searching the EMRs of the San Francisco and Palo Alto Veterans Affairs medical centers and randomly selecting 200 patients. To improve the diagnostic accuracy of the computational algorithm in cases where histopathological data are unavailable, we developed an index of suspicion to identify cases with a high index of suspicion for sarcoidosis (confirmed and probable) based on clinical and radiographic features alone using the American Thoracic Society practice guideline. Through medical record review, we determined the positive predictive value (PPV) of diagnosing sarcoidosis by two computational methods: using ICD codes alone and using ICD codes plus the high index of suspicion.

RESULTS

Among the 200 patients, 158 (79%) had a high index of suspicion for sarcoidosis. Of these 158 patients, 142 (89.9%) had documentation of nonnecrotizing granuloma, confirming biopsy-proven sarcoidosis. The PPV of using ICD codes alone was 79% (95% CI 78.6%-80.5%) for identifying sarcoidosis cases and 71% (95% CI 64.7%-77.3%) for identifying histopathologically confirmed sarcoidosis in the EMRs. The inclusion of the generated high index of suspicion to identify confirmed sarcoidosis cases increased the PPV significantly to 100% (95% CI 96.5%-100%). Histopathology documentation alone was 90% sensitive compared with high index of suspicion.

CONCLUSIONS

ICD codes are reasonable classifiers for identifying sarcoidosis cases within EMRs with a PPV of 79%. Using a computational algorithm to capture index of suspicion data elements could significantly improve the case-identification accuracy.

Impact of Early COVID-19 Pandemic on Common Ophthalmic Procedures Volumes: A US Claims-Based Analysis

PURPOSE

The COVID-19 pandemic has had a profound effect on the delivery of healthcare in the United States and globally. The aim of this study was to evaluate the impact of COVID-19 on common ophthalmic procedure utilization and normalization to pre-pandemic daily rates.

METHODS

Leveraging a national database, Clinformatics™ DataMart (OptumInsight, Eden Prairie, MN), procedure frequencies and daily averages, defined by Current Procedural Terminology codes, of common elective and non-elective procedures within multiple ophthalmology sub-specialties were calculated. Interrupted time-series analysis with a Poisson regression model and smooth spline functions was used to model trends in pre-COVID-19 (January 1, 2018-February 29, 2020) and COVID-19 (March 1, 2020-June 30, 2020) periods.

RESULTS

Of 3,583,231 procedures in the study period, 339,607 occurred during the early COVID-19 time period. Anti-vascular endothelial growth factor injections (44,412 to 39,774, RR 1.01, CI 0.99-1.02; p = .212), retinal detachment repairs (1,290 to 1,086, RR 1.07, CI 0.99-1.15; p = .103), and glaucoma drainage implants/trabeculectomies (706 to 487, RR 0.93, CI 0.83-1.04; p = .200) remained stable. Cataract surgery (61,421 to 33,054, RR 0.77; CI 0.76-0.78; p < .001), laser peripheral iridotomy (1,875 to 890, RR 0.82, CI 0.76-0.88; p < .001), laser trabeculoplasty (2,680 to 1,753, RR 0.79, CI 0.74-0.84; p < .001), and blepharoplasty (1,522 to 797, RR 0.71, CI 0.66-0.77; p < .001) all declined significantly. All procedures except laser iridotomy returned to pre-COVID19 rates by June 2020.

CONCLUSION

Most ophthalmic procedures that significantly declined during the COVID-19 pandemic were elective procedures. Among these, the majority returned to 2019 daily averages by June 2020.

Does the magic of BERT apply to medical code assignment? A quantitative study

Unsupervised pretraining is an integral part of many natural language processing systems, and transfer learning with language models has achieved remarkable results in downstream tasks. In the clinical application of medical code assignment, diagnosis and procedure codes are inferred from lengthy clinical notes such as hospital discharge summaries. However, it is not clear if pretrained models are useful for medical code prediction without further architecture engineering. This paper conducts a comprehensive quantitative analysis of various contextualized language models' performances, pretrained in different domains, for medical code assignment from clinical notes. We propose a hierarchical fine-tuning architecture to capture interactions between distant words and adopt label-wise attention to exploit label information. Contrary to current trends, we demonstrate that a carefully trained classical CNN outperforms attention-based models on a MIMIC-III subset with frequent codes. Our empirical findings suggest directions for building robust medical code assignment models.

Analysis of root causes of problems affecting the quality of hospital administrative data: A systematic review and Ishikawa diagram

INTRODUCTION

Administrative hospital databases represent an important tool for hospital financing in many national health systems and are also an important data source for clinical, epidemiological and health services research. Therefore, the data quality of such databases is of utmost importance. This paper aims to present a systematic review of root causes of data quality problems affecting administrative hospital data, creating a catalogue of potential issues for data quality analysts to explore.

METHODS

The MEDLINE and Scopus databases were searched using inclusion criteria based on two following concept blocks: (1) administrative hospital databases and (2) data quality. Studies' titles and abstracts were screened by two reviewers independently. Three researchers independently selected the screened studies based on their full texts and then extracted the potential root causes inferred from them. These were subsequently classified according to the Ishikawa model based on 6 categories: "Personnel", "Material", "Method", "Machine", "Mission" and "Management".

RESULTS

The result of our investigation and the contribution of this paper is a classification of the potential (105) root causes found through a systematic review of the 77 relevant studies we have identified and analyzed. The result was represented by an Ishikawa diagram. Most of the root causes (25.7%) were associated with the category "Personnel" - people's knowledge, preferences, education and culture, mostly related to clinical coders and health care providers activities. The quality of hospital documentation, within category "Material", and aspects related to financial incentives or disincentives, within category "Mission", were also frequently cited in the literature as relevant root causes for data quality issues.

CONCLUSIONS

The resultant catalogue of root causes, systematized using the Ishikawa framework, provides a compilation of potential root causes of data quality issues to be considered prior to reusing these data and that can point to actions aimed at improving data quality.

What Oncologists Want: Identifying Challenges and Preferences on Diagnosis Data Entry to Reduce EHR-Induced Burden and Improve Clinical Data Quality

PURPOSE

Accurate recording of diagnosis (DX) data in electronic health records (EHRs) is important for clinical practice and learning health care. Previous studies show statistically stable patterns of data entry in EHRs that contribute to inaccurate DX, likely because of a lack of data entry support. We conducted qualitative research to characterize the preferences of oncological care providers on cancer DX data entry in EHRs during clinical practice.

METHODS

We conducted semistructured interviews and focus groups to uncover common themes on DX data entry preferences and barriers to accurate DX recording. Then, we developed a survey questionnaire sent to a cohort of oncologists to verify the generalizability of our initial findings. We constrained our participants to a single specialty and institution to ensure similar clinical backgrounds and clinical experience with a single EHR system.

RESULTS

A total of 12 neuro-oncologists and thoracic oncologists were involved in the interviews and focus groups. The survey developed from these two initial thrusts was distributed to 19 participants yielding a 94.7% survey response rate. Clinicians reported similar user interface experiences, barriers, and dissatisfaction with current DX entry systems including repetitive entry operations, difficulty in finding specific DX options, time-consuming interactions, and the need for workarounds to maintain efficiency. The survey revealed inefficient DX search interfaces and challenging entry processes as core barriers.

CONCLUSION

Oncologists seem to be divided between specific DX data entry and time efficiency because of current interfaces and feel hindered by the burdensome and repetitive nature of EHR data entry. Oncologists' top concern for adopting data entry support interventions is ensuring that it provides significant time-saving benefits and increasing workflow efficiency. Future interventions should account for time efficiency, beyond ensuring data entry effectiveness.

Natural Language Processing in Surgery: A Systematic Review and Meta-analysis

OBJECTIVE

The aim of this study was to systematically assess the application and potential benefits of natural language processing (NLP) in surgical outcomes research.

SUMMARY BACKGROUND DATA

Widespread implementation of electronic health records (EHRs) has generated a massive patient data source. Traditional methods of data capture, such as billing codes and/or manual review of free-text narratives in EHRs, are highly labor-intensive, costly, subjective, and potentially prone to bias.

METHODS

A literature search of PubMed, MEDLINE, Web of Science, and Embase identified all articles published starting in 2000 that used NLP models to assess perioperative surgical outcomes. Evaluation metrics of NLP systems were assessed by means of pooled analysis and meta-analysis. Qualitative synthesis was carried out to assess the results and risk of bias on outcomes.

RESULTS

The present study included 29 articles, with over half (n = 15) published after 2018. The most common outcome identified using NLP was postoperative complications (n = 14). Compared to traditional non-NLP models, NLP models identified postoperative complications with higher sensitivity [0.92 (0.87-0.95) vs 0.58 (0.33-0.79), P < 0.001]. The specificities were comparable at 0.99 (0.96-1.00) and 0.98 (0.95-0.99), respectively. Using summary of likelihood ratio matrices, traditional non-NLP models have clinical utility for confirming documentation of outcomes/diagnoses, whereas NLP models may be reliably utilized for both confirming and ruling out documentation of outcomes/diagnoses.

CONCLUSIONS

NLP usage to extract a range of surgical outcomes, particularly postoperative complications, is accelerating across disciplines and areas of clinical outcomes research. NLP and traditional non-NLP approaches demonstrate similar performance measures, but NLP is superior in ruling out documentation of surgical outcomes.

Economic Implications of Chinese Diagnosis-Related Group-Based Payment Systems for Critically Ill Patients in ICUs

OBJECTIVES

To evaluate the economic implications of payments based on Chinese diagnosis-related groups for critically ill patients in ICUs in terms of total hospital expenditure, out-of-pocket payments, and length of stay.

DESIGN

A pre-post comparison of patient cohorts admitted to ICUs 1 year before and 1 year after Chinese diagnosis-related group reform was undertaken. Demographic characteristics, clinical data, and medical expenditures were collated from a health insurance database.

SETTING

Twenty-two public hospitals in Sanming, Southern China.

PATIENTS

All patients admitted to ICUs from January 1, 2017, to December 31, 2018.

INTERVENTION

The implementation of Chinese diagnosis-related group-based payments on January 1, 2018.

MEASUREMENTS AND MAIN RESULTS

Economic variables (total expenditures, out-of-pocket payments, and length of stay) were calculated for each patient from the day of hospital admission to the day of hospital discharge. Adjusted mean out-of-pocket payment estimates were 29.46% (p < 0.001) lower following reform. Adjusted mean out-of-pocket payments fell by 41.32% for patients in neonatal ICU, whereas there were no significant decreases in out-of-pocket payments for patients in PICU and adult ICU. Furthermore, adjusted mean out-of-pocket payments decreased by 55.74% in secondary hospitals, but there was no significant change in tertiary hospitals after Chinese diagnosis-related group reform. No significant changes were found in total expenditures and length of stay.

CONCLUSIONS

Chinese diagnosis-related group policy provided an opportunity for critically ill patients in ICUs to achieve at least short-term financial benefits in reducing out-of-pocket payments, without affecting the total expenditures and length of stay. Chinese diagnosis-related group-based payment significantly relieved financial burdens for patients with lower illness severities, such as patients in neonatal ICU. The results of this study can offer significant insights for policymakers in reducing the financial burden on critically ill patients, both in China and in other countries with similar systems.

From downcoding to upcoding: DRG based payment in hospitals

A prospective disease group-based payment is a reimbursement rule used in a wide array of countries. It turns to be the hospital's payment rule to imply. The secret of this payment is a fee payment as well as a hospital's activity based payment. There is a consensus to consider this rule of payment as the least likely to be manipulated by the actors. However, the defined fee per group depends on recorded information that is then processed using complex algorithms. What if the data itself can be manipulated? The result would be a fee per group based on manipulated factors that would lead to an inefficient budget allocation between hospitals. Using a unique French longitudinal database with 145 million stays, I unambiguously demonstrate that the implementation of a finer classification led to an upcoding-learning effect. The end result has been a budget transfer from public non-research hospitals to for-profit hospitals. The 2009 policy lead to upcoding disconnected from any changes in the trend of production of care.

Value of the International Classification of Diseases code for identifying children with biliary atresia

BACKGROUND

Although identifying cases in large administrative databases may aid future research studies, previous reports demonstrated that the use of the International Classification of Diseases, Tenth Revision (ICD-10) code alone for diagnosis leads to disease misclassification.

PURPOSE

We aimed to assess the value of the ICD-10 diagnostic code for identifying potential children with biliary atresia.

METHODS

Patients aged <18 years assigned the ICD-10 code of biliary atresia (Q44.2) between January 1996 and December 2016 at a quaternary care teaching hospital were identified. We also reviewed patients with other diagnoses of code-defined cirrhosis to identify more potential cases of biliary atresia. A proposed diagnostic algorithm was used to define ICD-10 code accuracy, sensitivity, and specificity.

RESULTS

We reviewed the medical records of 155 patients with ICD-10 code Q44.2 and 69 patients with other codes for biliary cirrhosis (K74.4, K74.5, K74.6). The accuracy for identifying definite/probable/possible biliary atresia cases was 80%, while the sensitivity was 88% (95% confidence interval [CI], 82%-93%). Three independent predictors were associated with algorithm-defined definite/probable/possible cases of biliary atresia: ICD-10 code Q44.2 (odds ratio [OR], 2.90; 95% CI, 1.09-7.71), history of pale stool (OR, 2.78; 95% CI, 1.18-6.60), and a presumed diagnosis of biliary atresia prior to referral to our hospital (OR, 17.49; 95% CI, 7.01-43.64). A significant interaction was noted between ICD-10 code Q44.2 and a history of pale stool (P<0.05). The area under the curve was 0.87 (95% CI, 0.84-0.89).

CONCLUSION

ICD-10 code Q44.2 has an acceptable value for diagnosing biliary atresia. Incorporating clinical data improves the case identification. The use of this proposed diagnostic algorithm to examine data from administrative databases may facilitate appropriate health care allocation and aid future research investigations.

New Index for Multiple Chronic Conditions Predicts Functional Outcome in Ischemic Stroke

OBJECTIVE

To determine whether a new index for multiple chronic conditions (MCCs) predicts poststroke functional outcome (FO), we developed and internally validated the new MCC index in patients with ischemic stroke.

METHODS

A prospective cohort of patients with ischemic stroke (2008-2017) was interviewed at baseline and 90 days in the Brain Attack Surveillance in Corpus Christi Project. An average of 22 activities of daily living (ADL)/instrumental ADL (IADL) items measured the FO score (range 1-4) at 90 days. A FO score >3 (representing a lot of difficulty with ADL/IADLs) was considered unfavorable FO. A new index was developed using machine learning techniques to select and weight conditions and prestroke impairments.

RESULTS

Prestroke modified Rankin Scale (mRS) score, age, congestive heart failure (CHF), weight loss, diabetes, other neurologic disorders, and synergistic effects (dementia × age, CHF × renal failure, and prestroke mRS × prior stroke/TIA) were identified as important predictors in the MCC index. In the validation dataset, the index alone explained 31% of the variability in the FO score, was well-calibrated (p = 0.41), predicted unfavorable FO well (area under the receiver operating characteristic curve 0.81), and outperformed the modified Charlson Comorbidity Index in predicting the FO score and poststroke mRS.

CONCLUSIONS

A new MCC index was developed and internally validated to improve the prediction of poststroke FO. Novel predictors and synergistic interactions were identified.

CLASSIFICATION OF EVIDENCE

This study provides Class II evidence that in patients with ischemic stroke, an index for MCC predicts FO at 90 days.

Challenges and Adverse Outcomes of Implementing Reimbursement Mechanisms Based on the Diagnosis-Related Group Classification System: A systematic review

In health insurance, a reimbursement mechanism refers to a method of third-party repayment to offset the use of medical services and equipment. This systematic review aimed to identify challenges and adverse outcomes generated by the implementation of reimbursement mechanisms based on the diagnosis-related group (DRG) classification system. All articles published between 1983 and 2017 and indexed in various databases were reviewed. Of the 1,475 articles identified, 36 were relevant and were included in the analysis. Overall, the most frequent challenges were increased costs (especially for severe diseases and specialised services), a lack of adequate supervision and technical infrastructure and the complexity of the method. Adverse outcomes included reduced length of patient stay, early patient discharge, decreased admissions, increased re-admissions and reduced services. Moreover, DRG-based reimbursement mechanisms often resulted in the referral of patients to other institutions, thus transferring costs to other sectors.

How much can we trust electronic health record data?

Trust in EHR data is becoming increasingly important as a greater share of clinical and health services research use EHR data. We discuss reasons for distrust and acknowledge limitations. Researchers continue to use EHR data because of strengths including greater clinical detail than sources like administrative billing claims. Further, many limitations are addressable with existing methods including data quality checks and common data frameworks. We discuss how to build greater trust in the use of EHR data for research, including additional transparency and research priority areas that will both enhance existing strengths of the EHR and mitigate its limitations.

Workflow Differences Affect Data Accuracy in Oncologic EHRs: A First Step Toward Detangling the Diagnosis Data Babel

PURPOSE

Diagnosis (DX) information is key to clinical data reuse, yet accessible structured DX data often lack accuracy. Previous research hints at workflow differences in cancer DX entry, but their link to clinical data quality is unclear. We hypothesized that there is a statistically significant relationship between workflow-describing variables and DX data quality.

METHODS

We extracted DX data from encounter and order tables within our electronic health records (EHRs) for a cohort of patients with confirmed brain neoplasms. We built and optimized logistic regressions to predict the odds of fully accurate (ie, correct neoplasm type and anatomic site), inaccurate, and suboptimal (ie, vague) DX entry across clinical workflows. We selected our variables based on correlation strength of each outcome variable.

RESULTS

Both workflow and personnel variables were predictive of DX data quality. For example, a DX entered in departments other than oncology had up to 2.89 times higher odds of being accurate (P < .0001) compared with an oncology department; an outpatient care location had up to 98% fewer odds of being inaccurate (P < .0001), but had 458 times higher odds of being suboptimal (P < .0001) compared with main campus, including the cancer center; and a DX recoded by a physician assistant had 85% fewer odds of being suboptimal (P = .005) compared with those entered by physicians.

CONCLUSION

These results suggest that differences across clinical workflows and the clinical personnel producing EHR data affect clinical data quality. They also suggest that the need for specific structured DX data recording varies across clinical workflows and may be dependent on clinical information needs. Clinicians and researchers reusing oncologic data should consider such heterogeneity when conducting secondary analyses of EHR data.

Big Data vs. Clinical Trials in HPB Surgery

Randomized controlled clinical trials (RCTs) are at the heart of "evidence-based" medicine. However, in surgical practice, RCTs remain uncommon. Conducting well-designed RCTs for surgical procedures is often challenged by inadequate recruitment accrual, blinding, or standardization of the surgical procedure, as well as lack of funding and evolution of the treatment strategy during the many years over which such trials are conducted. In addition, most clinical trials are performed in academic high-volume centers in highly selected patients, which may not necessarily reflect a "real-world" practice setting. Over the past decades, surgical outcomes research using nationwide administrative and registry databases has become increasingly common. Large databases provide easy and inexpensive access to data on a large and diverse patient population at a variety of treatment centers. Furthermore, large database studies provide the opportunity to answer questions that would be impossible or very arduous to answer using RCTs, including questions regarding health policy efficacy, trends in surgical practice, access to health care, impact of hospital volume, and adherence to practice guidelines, as well as research questions regarding rare disease, infrequent surgical outcomes, and specific subpopulation. Prospective data registries may also allow for quality benchmarking and auditing. This review outlines the role, advantages, and limitations of RCTs and large database studies in answering important research questions in surgery.

Physical and Occupational Therapy Use and Cost After Common Hand Procedures

PURPOSE

The use of routine physical therapy (PT) and occupational therapy (OT) after certain hand procedures, such as carpal tunnel release, remains controversial. The objective of this study was to evaluate baseline use, the change in use, variation in prescribing patterns by region, and costs for PT/OT after common hand procedures.

METHODS

Outpatient administrative claims data from patients who underwent procedures for carpal tunnel syndrome, trigger finger, carpometacarpal arthritis, de Quervain tenosynovitis, wrist ganglion cyst, and distal radius fracture were abstracted from the Truven Health MarketScan database from 2007 to 2015. The incidence of therapy and total reimbursement of therapy per patient were collected for each procedure over a 90-day postoperative observational period. Trends in use of therapy over time were described with average compound annual growth rates (CAGRs), a way of quantifying average growth over a specified observation period. Variations in the incidence of PT/OT use across 4 census regions were assessed.

RESULTS

The incidence of 90-day utilization of PT and OT after hand procedures was 14.0% and increased for all procedures during the observation period with an average CAGR of 8.3%. Cost per therapy visit was relatively stable when adjusted for inflation, with an average CAGR of 0.63%. Patients in the northeast had a significantly higher incidence of PT/OT use than those in the south and west for all procedures except carpometacarpal arthritis.

CONCLUSIONS

Use of PT and OT has increased over time after common hand procedures. Geographical variation in the utilization rate of these services is substantial. Limiting unwarranted variation of care is a health policy strategy for increasing value of care.

TYPE OF STUDY/LEVEL OF EVIDENCE

Outcomes Research II.

Comparing the Performance of Charlson and Elixhauser Comorbidity Indices to Predict In-Hospital Mortality Among a Chinese Population

Objective

Earlier comorbidity measures have been developed or validated using the North American population. This study aims to compare five Charlson or Elixhauser comorbidity indices to predict in-hospital mortality using a large electronic medical record database from Shanxi, China.

Methods

Using the primary diagnosis code and surgery procedure codes, we identified four hospitalized patient cohorts, hospitalized between 2013 and 2017, in Shanxi, China, as follows: congestive heart failure (CHF, n=41,577), chronic renal failure (CRF, n=40,419), diabetes (n=171,355), and percutaneous coronary intervention (PCI, n=39,097). We used logistic regression models and c-statistics to evaluate the in-hospital mortality predictive performance of two multiple comorbidity indicator variables developed by Charlson in 1987 and Elixhauser in 1998 and three single numeric scores by Quan in 2011, van Walraven in 2009, and Moore 2017.

Results

Elixhauser comorbidity indicator variables had consistently higher c-statistics (0.824, 0.843, 0.904, 0.853) than all other four comorbidity measures, across all four disease cohorts. Moore’s comorbidity score outperformed the other two score systems in CHF, CRF, and diabetes cohorts (c-statistics: 0.776, 0.832, 0.869), while van Walraven’s score outperformed all others among PCI patients (c-statistics: 0.827).

Conclusion

Elixhauser comorbidity indicator variables are recommended, when applied to large Chinese electronic medical record databases, while Moore’s score system is appropriate for relatively small databases.

Physician code creep after the initiation of outpatient volume control program and implications for appropriate ICD-10-CM coding

BACKGROUND

Most studies on the physician code creep (i.e., changes in case mix record-keeping practices to improve reimbursement) have focused on episodes (inpatient hospitalizations or outpatient procedures). Little is known regarding changes in diagnostic coding practices for better reimbursement among a fixed cohort of patients with chronic diseases.

METHODS

To examine whether physicians in tertiary medical centers changed their coding practices after the initiation of the Outpatient Volume Control Program (OVCP) in Taiwan, we conducted a retrospective observational study of four patient cohorts (two interventions and two controls) from January 2016 to September 2017 in Taiwan. The main outcomes were the number of outpatient visits with four coding practices: 1) OVCP monitoring code recorded as primary diagnosis; 2) OVCP monitoring code recorded as secondary diagnosis; 3) non-OVCP monitoring code recorded as primary diagnosis; 4) non-OVCP monitoring code recorded as secondary diagnosis.

RESULTS

The percentage change of the number of visits with coding practice 1 between 2016Q1 and 2017Q3 was - 74% for patients with hypertension and - 73% with diabetes in tertiary medical centers and - 23% and - 17% in clinics, respectively. By contrast, the percentage changes of coding practice 3 were + 73% for patients with hypertension and + 46% for patients with diabetes in tertiary medical centers and - 19% and - 2% in clinics, respectively.

CONCLUSIONS

Physician code creep occurred after the initiation of the OVCP. Education regarding appropriate outpatient coding for physicians will be relatively effective when proper coding is related to reimbursement.

Validation of new ICD-10-based patient safety indicators for identification of in-hospital complications in surgical patients: a study of diagnostic accuracy

OBJECTIVE

Administrative data systems are used to identify hospital-based patient safety events; few studies evaluate their accuracy. We assessed the accuracy of a new set of patient safety indicators (PSIs; designed to identify in hospital complications).

STUDY DESIGN

Prospectively defined analysis of registry data (1 April 2010-29 February 2016) in a Canadian hospital network. Assignment of complications was by two methods independently. The National Surgical Quality Improvement Programme (NSQIP) database was the clinical reference standard (primary outcome=any in-hospital NSQIP complication); PSI clusters were assigned using International Classification of Disease (ICD-10) codes in the discharge abstract. Our primary analysis assessed the accuracy of any PSI condition compared with any complication in the NSQIP; secondary analysis evaluated accuracy of complication-specific PSIs.

PATIENTS

All inpatient surgical cases captured in NSQIP data.

ANALYSIS

We assessed the accuracy of PSIs (with NSQIP as reference standard) using positive and negative predictive values (PPV/NPV), as well as positive and negative likelihood ratios (±LR).

RESULTS

We identified 12 898 linked episodes of care. Complications were identified by PSIs and NSQIP in 2415 (18.7%) and 2885 (22.4%) episodes, respectively. The presence of any PSI code had a PPV of 0.55 (95% CI 0.53 to 0.57) and NPV of 0.93 (95% CI 0.92 to 0.93); +LR 6.41 (95% CI 6.01 to 6.84) and -LR 0.40 (95% CI 0.37 to 0.42). Subgroup analyses (by surgery type and urgency) showed similar performance. Complication-specific PSIs had high NPVs (95% CI 0.92 to 0.99), but low to moderate PPVs (0.13-0.61).

CONCLUSION

Validation of the ICD-10 PSI system suggests applicability as a first screening step, integrated with data from other sources, to produce an adverse event detection pathway that informs learning healthcare systems. However, accuracy was insufficient to directly identify or rule out individual-level complications.

Factors associated with high-cost hospitalization for peritonitis in children receiving chronic peritoneal dialysis in the United States

BACKGROUND

Although peritonitis causes significant morbidity and mortality in children receiving chronic peritoneal dialysis (CPD), little is known about costs associated with treatment.

METHODS

We analyzed 246 peritonitis-related hospitalizations in the USA, linked by the Standardized Care to Improve Outcomes in Pediatric End Stage Renal Disease (SCOPE) and Pediatric Health Information Systems (PHIS) databases. Multivariable logistic regression was used to assess the relationship between high-cost hospitalizations (at or above the 75th percentile) and patient characteristics. Multivariable modeling was used to assess differences in the service-line specific geometric mean between (1) high- and low-cost (below the 75th percentile) hospitalizations and (2) fungal versus other types of peritonitis. Wage-adjusted hospitalization charges were converted to estimated costs using reported cost-to-charge ratios to estimate the cost of hospitalization.

RESULTS

High-cost hospitalizations were associated with the following: age 3-12 years, Hispanic ethnicity, intensive care unit (ICU) stay, length of stay (LOS), and fungal peritonitis. Whereas absolute standardized cost by service line was significantly different when comparing high- and low-cost hospitalizations, the percentage of total cost by service line was similar in the two groups. Cost per case for fungal peritonitis was higher (p < 0.001) in every service line except pharmacy when compared to other peritonitis cases. The median (IQR) cost of hospitalization for the treatment of peritonitis was $13,655 ($7871, $28434) USD.

CONCLUSIONS

Hospitalization-related costs for peritonitis treatment are substantial and arise from a variety of service lines. Fungal peritonitis is associated with high-cost hospitalization.

Lost in Translation: Diagnosis Records Show More Inaccuracies After Biopsy in Oncology Care EHRs

The use of diagnosis (DX) data is crucial to secondary use of electronic health record (EHR) data, yet accessible structured DX data often lack in accuracy. DX descriptions associated with structured DX codes vary even after recording biopsy results; this may indicate poor data quality. We hypothesized that biopsy reports in cancer care charts do not improve intrinsic DX data quality. We analyzed DX data for a manually well-annotated cohort of patients with brain neoplasms. We built statistical models to predict the number of fully-accurate (i.e., correct neoplasm type and anatomical location) and inaccurate DX (i.e. type or location contradicts cohort data) descriptions. We found some evidence of statistically larger numbers of fully-accurate (RR=3.07, p=0.030) but stronger evidence of much larger numbers of inaccurate DX (RR=12.3, p=0.001 and RR=19.6, p<0.0001) after biopsy result recording. Still, 65.9% of all DX records were neither fully-accurate nor fully-inaccurate. These results suggest EHRs must be modified to support more reliable DX data recording and secondary use of EHR data.

Using SNOMED CT-encoded problems to improve ICD-10-CM coding-A randomized controlled experiment

OBJECTIVE

Clinical problems in the Electronic Health Record that are encoded in SNOMED CT can be translated into ICD-10-CM codes through the NLM's SNOMED CT to ICD-10-CM map (NLM Map). This study evaluates the potential benefits of using the map-generated codes to assist manual ICD-10-CM coding.

METHODS

De-identified clinic notes taken by the physician during an outpatient encounter were made available on a secure web server and randomly assigned for coding by professional coders with usual coding or map-assisted coding. Map-assisted coding made use of the problem list maintained by the physician and the NLM Map to suggest candidate ICD-10-CM codes to the coder. A gold standard set of codes for each note was established by the coders using a Delphi consensus process. Outcomes included coding time, coding reliability as measured by the Jaccard coefficients between codes from two coders with the same method of coding, and coding accuracy as measured by recall, precision and F-score according to the gold standard.

RESULTS

With map-assisted coding, the average coding time per note reduced by 1.5 min (p = 0.006). There was a small increase in coding reliability and accuracy (not statistical significant). The benefits were more pronounced in the more experienced than less experienced coders. Detailed analysis of cases in which the correct ICD-10-CM codes were not found by the NLM Map showed that most failures were related to omission in the problem list and suboptimal mapping of the problem list terms to SNOMED CT. Only 12% of the failures was caused by errors in the NLM Map.

CONCLUSION

Map-assisted coding reduces coding time and can potentially improve coding reliability and accuracy, especially for more experienced coders. More effort is needed to improve the accuracy of the map-suggested ICD-10-CM codes.

Infections and internal medicine patients: Could a comorbidity score predict in-hospital mortality?

Infectious diseases (ID) are frequently cause of internal medicine wards (IMW) admission. We aimed to evaluate risk factors for in-hospital mortality (IHM) in IMW patients with ID, and to test the usefulness of a comorbidity score (CS).This study included ID hospital admissions between January 2013, and December 2016, recorded in the database of the local hospital. ICD-9-CM codes were selected to identify infections, development of sepsis, and to calculate a CS.We analyzed 12,173 records, (age 64.8 ± 25.1 years, females 66.2%, sepsis 9.3%). Deceased subjects (1545, 12.7%) were older, had higher percentage of sepsis, pulmonary infections, and endocarditis. Mean value of CS was also significantly higher. At multivariate analysis, the odds ratio (OR) for sepsis (OR 5.961), endocarditis (OR 4.247), pulmonary infections (OR 1.905), other sites of infection (OR 1.671), and urinary tracts infections (OR 0.548), were independently associated with IHM. The CS (OR 1.070 per unit of increasing score), was independently associated with IHM as well. The calculated weighted risk, obtained by multiplying 1.070 for the mean score value in deceased patients, was 19.367. Receiver operating characteristic (ROC) analysis showed that CS and development of sepsis were significant predictors for IHM (area under the curve, AUC: 0.724 and 0.670, respectively).Careful evaluation of comorbidity in internal medicine patients is nowadays matter of extreme importance in IMW patients hospitalized for ID, being IHM related to severity of disease, type and site of infection, and also to concomitant comorbidities. In these patients, a careful evaluation of CS should represent a fundamental step in the disease management.

Incidence and disease burden of chemotherapy-induced peripheral neuropathy in a population-based cohort

OBJECTIVE

To assess disease burden of chemotherapy-induced peripheral neuropathy (CIPN), which is a common dose-limiting side effect of neurotoxic chemotherapy. Late effects of CIPN may increase with improved cancer survival.

METHODS

Olmsted County, Minnesota residents receiving neurotoxic chemotherapy were identified and CIPN was ascertained via text searches of polyneuropathy symptoms in the medical record. Clinical records were queried to collect data on baseline characteristics, risk factors, signs and symptoms of CIPN, medications, impairments and International Classification of Diseases, Ninth Revision (ICD-9) diagnostic codes for all subjects.

RESULTS

A total of 509 individuals with incident exposure to an inclusive list of neurotoxic chemotherapy agents between 2006 and 2008 were identified. 268 (52.7%) of these individuals were determined to have CIPN. The median time from incident exposure to first documented symptoms was 71 days. Patients with CIPN received a neuropathy ICD-9 diagnosis in only 37 instances (13.8%). Pain symptoms and use of pain medications were observed more often in patients with CIPN. Five-year survival was greater in those with CIPN (55.2%) versus those without (36.1%). Those with CIPN surviving greater than 5 years (n=145) continued to have substantial impairments and were more likely to be prescribed opioids than those without CIPN (OR 2.0, 1.06-3.69).

CONCLUSIONS

Results from our population-based study are consistent with previous reports of high incidence of CIPN in the first 2 years following incident exposure to neurotoxic chemotherapeutic agents, and its association with significant pain symptomatology and accompanied long-term opioid use. Increased survival following exposure to neurotoxic chemotherapy and its long-term disease burden necessitates further study among survivors.

Biopsy Records Do Not Reduce Diagnosis Variability in Cancer Patient EHRs: Are We More Uncertain After Knowing?

Diagnostic codes are crucial for analyses of electronic health record (EHR) data but their accuracy and precision are often lacking. Although providers enter precise diagnoses into progress notes, billing standards may limit the particularity of a diagnostic code. Variability also arises from the creation of multiple descriptions for a particular diagnostic code. We hypothesized that the variability of diagnostic codes would be greater before surgical pathology results were recorded in the medical record. A well annotated cohort of patients with brain neoplasms was studied. After diagnostic pathology reporting, the odds of more distinct diagnostic descriptions were 2.30 times higher (p=0.00358), entropy in diagnostic sequences was 2.26 times higher (p=0.0259) and entropy in diagnostic precision scores was 15.5 times higher (p=0.0324). Although diagnostic codes became more distinct on average after diagnostic pathology reporting, there was a paradoxical increase in the variability of the codes selected. Researchers must be aware of the inconsistencies and variability in particularity in structured diagnostic coding despite the presence of a definitive diagnosis.

Clinical and Administrative Data on the Research of Acute Coronary Syndrome in Spain. Minimum Basic Data Set Validity

INTRODUCTION AND OBJECTIVES

Health outcomes research is done from clinical registries or administrative databases. The aim of this work was to evaluate the concordance of the Minimum Basic Data Set (MBDS) with the DIOCLES (Descripción de la Cardiopatía Isquémica en el Territorio Español) registry and to analyze the implications of use of the MBDS in the study of acute coronary syndrome in Spain.

METHODS

Through indirect identifiers, DIOCLES was linked with MBDS and unique matches were selected. Some of most relevant variables for risk adjustment of in-hospital mortality due to acute myocardial infarction were considered. Kappa coefficient was used to evaluate the concordance; sensitivity, specificity and positive and negative predictive values to measure the validity of the MBDS, and the area under ROC (receiver operating characteristic) curve to calculate its discrimination. The results were compared among hospitals quintiles according to their contribution to DIOCLES. The influence of unmatched episodes on results was assessed by a sensitivity analysis, using looser linking criteria.

RESULTS

Overall, 1539 (60.85%) unique matches were achieved. The prevalence was higher in DIOCLES (acute myocardial infarction: 71.09%; Killip 3-4: 9.17%; cerebrovascular accident: 0.97%; thrombolysis: 8.64%; angioplasty: 61.92% and coronary bypass: 1.75%) than in the MBDS (P < .001). The agreement level observed was almost perfect (κ = 0.863). The MBDS showed a sensitivity of 85.10% and a specificity of 98.31%. Most results were confirmed by using sensitivity analysis (79.95% episodes matched).

CONCLUSIONS

The MBDS can be a useful tool for outcomes research of acute coronary syndrome in Spain. The contrast of DIOCLES and MBDS with medical records could verify their validity.

External fixation and surgical fusion for pediatric cervical spine injuries: Short-term outcomes

OBJECTIVE

To compare in-hospital complication rates in pediatric patients with atlantoaxial and subaxial injuries undergoing either external fixation or surgical fusion.

PATIENTS AND METHODS

Baseline and outcome data were obtained from the 2002-2011 Nationwide Inpatient Sample (NIS) for patients under the age of 18 with a diagnosis of cervical spine fracture without spinal cord injury or cervical spine subluxation. Patients who underwent external immobilization or internal fixation were included for analysis. Variables analyzed included length of stay, in-hospital mortality, discharge disposition, total hospital charges, and development of at least one in-hospital complication.

RESULTS

A total of 2878 pediatric patients with cervical spine injury were identified; 1462 patients (50.8%) with atlantoaxial (C1-2) injury and 1416 (49.2%) with subaxial (C3-7) injury. Among atlantoaxial injury patients, external fixation was associated with lower total charges ($73,786 vs. $98,158, p = .040) and a lower likelihood of developing at least one complication (1.9% vs. 6.8%, p = .029) compared to surgical fusion, and was a more common treatment for subluxation alone (16.4% vs. 2.6%, p < .001). Among subaxial injury patients, there were no significant differences in age (p = .262), length of stay (p = .196), occurrence of at least one complication (p = .334), or total charges (p = .142). Subaxial subluxation injuries alone were treated more often with surgical fusion (2.2% vs. 1.2%, p < .001).

CONCLUSION

Optimal treatment of patients with cervical injury may vary by location of injury. Our findings warrant further investigation into the difference in clinical outcomes between surgical and non-surgical management of atlantoaxial and subaxial injury.

Can hospital discharge data be used for monitoring indicators associated with postpartum hemorrhages? The HERA multicenter observational study

OBJECTIVES

The principal objective of this study was to assess the reliability of measuring the incidence of postpartum hemorrhages (PPH) from the national hospital discharge summary database (PMSI). The secondary objectives were to assess this reliability according to the maternity unit level and status and to assess the measurement of second-line procedures for PPH.

MATERIALS AND METHODS

This study compared PPH incidence rates from February through July 2011 in 131 maternity units, as measured in the PMSI and the prospective HERA study cohort, considered as the reference standard.

RESULTS

Compared with the cohort, PPH incidence was over-reported in the PMSI among vaginal deliveries (4.0% vs. 3.5; P<0.0001), but not cesareans (3.2 vs. 2.9%; P=0.1). For the second-line curative procedures, PMSI data underestimated the incidence of vessel embolization and transfusion (P<0.0001) among vaginal deliveries and of hypogastric ligation (P=0.002), other vessel ligation (P=0.005), and transfusion (P<0.0001) among cesareans.

CONCLUSION

Despite some coding inaccuracy in the PMSI, routinely collected data can provide acceptable estimates for maternity units and perinatal networks to use to improve quality of care through the monitoring of quality indicators. Improvements are nonetheless needed for international comparisons and other epidemiologic purposes.

Effectiveness of Umeclidinium-Vilanterol for Protocolized Management of Chronic Obstructive Pulmonary Disease Exacerbation in Hospitalized Patients: A Sequential Period Analysis

Background: Bronchodilator therapy is a foundation of chronic obstructive pulmonary disease (COPD) exacerbation treatment. Although international guidelines recommend short-acting formulations given multiple times per day, long-acting formulations have not been adequately evaluated. The objective of our study was to determine the effectiveness of umeclidinium-vilanterol (UME/VIL), long-acting beta2-agonist/long-acting muscarinic antagonist (LABA/LAMA) as a once-daily alternative for treating COPD exacerbations in hospitalized patients. Methods: In this retrospective sequential period analysis, we reviewed electronic medical records of patients hospitalized for COPD exacerbations before (September 1, 2015 to February 29, 2016) and after (April 1, 2016 to September 30, 2016) incorporation of UME/VIL into our standard COPD protocol. Before implementation, patients received a daily anticholinergic plus twice-daily long-acting beta2-agonist therapy (tiotropium plus formoterol, n=65). After implementation, UME/VIL replaced the previous regimen (n=58). No other changes were made to the COPD protocol. The primary outcome was 30-day hospital readmission rate. Hospital length of stay, 30-day mortality, and cost of care were analyzed as secondary outcomes. Results: A trend toward increased 30-day readmission rates in the post-intervention group (24.1% versus 10.8%, p=0.049) was no longer statistically significant after adjustment for severity of illness (based on case-mix index) and complications or comorbidities based on diagnosis-related group codes (adjusted odds ratio: 2.499; 95% confidence interval: 0.916-7.380; p=0.074). Conclusion: After adjustment for potential confounders,the implementation of a LABA/LAMA combination product was not statistically associated with an increased 30-day readmission rate but was associated with lower cost of care.

Racial Differences in Mortality from Severe Acute Respiratory Failure in the United States, 2008-2012

RATIONALE

Racial disparities in health and healthcare in the United States are well documented and are increasingly recognized in acute critical illnesses such as sepsis and acute respiratory failure.

OBJECTIVES

Using a large, representative, U.S. nationwide database, we examined the hypothesis that black and Hispanic patients with severe acute respiratory failure have higher mortality rates when compared with non-Hispanic whites.

METHODS

This retrospective analysis used discharge data from the Agency for Healthcare Research and Quality, Nationwide Inpatient Sample, Healthcare Cost and Utilization Project, for the years 2008-2012. We identified hospitalizations with acute respiratory failure using a combination of International Classification of Diseases, Ninth Revision, Clinical Modification, codes. A logistic regression model was fitted to compare in-hospital mortality rates by race.

MEASUREMENTS AND MAIN RESULTS

After adjusting for sex, age, race, disease severity, type of hospital, and median household income for patient ZIP code, blacks had a greater odds ratio of in-hospital death when compared with non-Hispanic whites (odds ratio [OR], 1.13; 95% confidence interval [CI], 1.11-1.14; P < 0.001), and Hispanics also had a greater odds ratio of in-hospital death when compared with non-Hispanic whites (OR, 1.17; 95% CI, 1.15-1.19; P < 0.001), and so did Asian and Pacific Islanders (OR, 1.15; 95% CI, 1.12-1.18; P < 0.001) and Native Americans (OR, 1.08; 95% CI, 1.00-1.15; P < 0.001) when compared with non-Hispanic whites (OR, 1.0).

CONCLUSIONS

Blacks, Hispanics, and other racial minorities in the United States were observed to exhibit significantly higher in-hospital sepsis-related respiratory failure associated mortality when compared with non-Hispanic whites.

Classifying Chronic Lower Respiratory Disease Events in Epidemiologic Cohort Studies

RATIONALE

One in 12 adults has chronic obstructive pulmonary disease or asthma. Acute exacerbations of these chronic lower respiratory diseases (CLRDs) are a major cause of morbidity and mortality. Valid approaches to classifying cases and exacerbations in the general population are needed to facilitate prevention research.

OBJECTIVES

To assess the feasibility, reproducibility, and performance of a protocol to identify CLRD cases and exacerbations triggering emergency department (ED) visits or hospitalizations in cohorts of patients derived from general populations of adults.

METHODS

A protocol was developed to classify CLRD cases and severe exacerbations on the basis of review of medical records. ED and inpatient medical records were ascertained prospectively in the Hispanic Community Health Study/Study of Latinos, and inpatient records were retrospectively identified by administrative codes in the Multi-Ethnic Study of Atherosclerosis. "Probable" exacerbations were defined as a physician's diagnosis of CLRD with acute respiratory symptoms. "Highly probable" exacerbations additionally required systemic corticosteroid therapy, and "definite" exacerbations required airflow limitation or evidence of CLRD on imaging studies. Adjudicated results were compared with CLRD cases identified by spirometry and self-report, and with an administrative definition of exacerbations.

MEASUREMENTS AND MAIN RESULTS

Protocol-based classification was completed independently by two physicians for 216 medical records (56 ED visits and 61 hospitalizations in the Hispanic Community Health Study/Study of Latinos; 99 hospitalizations in the Multi-Ethnic Study of Atherosclerosis). Reviewer disagreement occurred in 2-5% of cases and 4-8% of exacerbations. Eighty-nine percent of records were confirmed as at least probable CLRD cases. Fifty-six percent of confirmed CLRD cases had airflow limitation on the basis of baseline study spirometry. Of records that described CLRD as the primary discharge diagnosis code, an acute exacerbation was confirmed as at least probable for 96% and as highly probable or definite for 77%. Only 50% of records with CLRD as a secondary code were confirmed, although such records accounted for over half of all confirmed exacerbations.

CONCLUSIONS

CLRD cases and severe exacerbations without preceding documentation of airflow limitation are identified frequently in population-based cohorts of persons. A primary discharge diagnosis of CLRD is specific but insensitive for defining exacerbations. Protocol-based classification of medical records may be appropriate to supplement and to validate identification of CLRD cases and exacerbations in general population studies. Clinical trials registered with www.clinicaltrials.gov (NCT00005487 and NCT02060344).

Accuracy of Billing Codes Used in the Therapeutic Care of Diabetic Retinopathy

Importance

Insurance billing claim databases represent a growing field of scientific inquiry within ophthalmology. Validating the accuracy of billing claim codes used during the care of diabetic retinopathy is a necessary precursor to fully understanding the underlying data and subsequent results of these types of studies.

Objective

To determine the accuracy of diagnostic, procedural, and therapeutic billing codes used in the treatment of diabetic retinopathy.

Design, Setting, and Participants

This retrospective medical record review was conducted at 3 clinical practices (1 academic and 2 private). Insured patients with diabetic retinopathy were seen by the practices between 2011 and 2013. Each patient then had every visit for 2 years reviewed twice, once for billing data and the second for data from the medical record. Data were collected and analyzed from October 2015 to July 2016.

Main Outcomes and Measures

The positive predictive value (PPV) and negative predictive value (NPV) for each code of interest. Sensitivity and specificity were secondary outcomes.

Results

A total of 146 patients (mean [SD] age, 60.3 [12.5] years) from 11 physicians had 1072 encounters reviewed over 2 calendar years. Among the included patients, 49.3% were female (n = 72), 48.6% were white (n = 71), 37.0% were black (n = 54), and 18.5% had type 1 diabetes and a mean (SD) hemoglobin A1C level of 7.7% (1.8) (n = 27). Nearly all codes of interest that were used frequently also had a high PPV (range, 89.5%-100%) and NPV (88.6%-100%) including billing codes for intravitreal injection, focal laser, panretinal photocoagulation, laterality of procedure, ranibizumab, bevacizumab, fundus photographs, fluorescein angiography, and optical coherence tomography. Codes that were used infrequently (<20 instances) but still had a high PPV (all 100%) and NPV (99.7%-100%) were codes for aflibercept, triamcinolone, and the dexamethasone implant. Only the codes for infrequently used B-scan ultrasonography (PPV, 69.6%) and subtenon injection (PPV, 100%; NPV, 99.7%, but sensitivity of only 40%) were found to be of questionable accuracy. Other than subtenon injection (40%), all codes were also found to have a high sensitivity (range, 87.6%-100%) and a high specificity (range, 97.2%-100%).

Conclusions and Relevance

These data suggest diagnostic, procedure, and therapeutic codes derived from insurance billing claims accurately reflect the medical record for patients with diabetic retinopathy.

The Validity of Discharge Billing Codes Reflecting Severe Maternal Morbidity

BACKGROUND

Discharge diagnoses are used to track national trends and patterns of maternal morbidity. There are few data regarding the validity of the International Classification of Diseases (ICD) codes used for this purpose. The goal of our study was to try to better understand the validity of administrative data being used to monitor and assess trends in morbidity.

METHODS

Hospital stay billing records were queried to identify all delivery admissions at the Massachusetts General Hospital for the time period 2001 to 2011 and the University of Michigan Health System for the time period 2005 to 2011. From this, we identified patients with ICD-9-Clinical Modification (CM) diagnosis and procedure codes indicative of severe maternal morbidity. Each patient was classified with 1 of 18 different medical/obstetric categories (conditions or procedures) based on the ICD-9-CM code that was recorded. Within each category, 20 patients from each institution were selected at random, and the corresponding medical charts were reviewed to determine whether the ICD-9-CM code was assigned correctly. The percentage of correct codes for each of 18 preselected clinical categories was calculated yielding a positive predictive value (PPV) and 99% confidence interval (CI).

RESULTS

The overall number of correctly assigned ICD-9-CM codes, or PPV, was 218 of 255 (86%; CI, 79%-90%) and 154 of 188 (82%; CI, 74%-88%) at Massachusetts General Hospital and University of Michigan Health System, respectively (combined PPV, 372/443 [84%; CI, 79-88%]). Codes within 4 categories (Hysterectomy, Pulmonary edema, Disorders of fluid, electrolyte and acid-base balance, and Sepsis) had a 99% lower confidence limit ≥75%. Codes within 8 additional categories demonstrated a 99% lower confidence limit between 74% and 50% (Acute respiratory distress, Ventilation, Other complications of obstetric surgery, Disorders of coagulation, Cardiomonitoring, Acute renal failure, Thromboembolism, and Shock). Codes within 6 clinical categories demonstrated a 99% lower confidence limit <50% (Puerperal cerebrovascular disorders, Conversion of cardiac rhythm, Acute heart failure [includes arrest and fibrillation], Eclampsia, Neurotrauma, and Severe anesthesia complications).

CONCLUSIONS

ICD-9-CM codes capturing severe maternal morbidity during delivery hospitalization demonstrate a range of PPVs. The PPV was high when objective supportive evidence, such as laboratory values or procedure documentation supported the ICD-9-CM code. The PPV was low when greater judgment, interpretation, and synthesis of the clinical data (signs and symptoms) was required to support a code, such as with the category Severe anesthesia complications. As a result, these codes should be used for administrative research with more caution compared with codes primarily defined by objective data.

Manual Compression versus Vascular Closing Device for Closing Access Puncture Site in Femoral Left-Heart Catheterization and Percutaneous Coronary Interventions: A Retrospective Cross-Sectional Comparison of Costs and Effects in Inpatient Care

OBJECTIVES

To compare complication rates, length of hospital stay, and resulting costs between the use of manual compression and a vascular closing device (VCD) in both diagnostic and interventional catheterization in a German university hospital setting.

METHODS

A stratified analysis according to risk profiles was used to compare the risk of complications in a retrospective cross-sectional single-center study. Differences in costs and length of hospital stay were calculated using the recycled predictions method, based on regression coefficients from generalized linear models with gamma distribution. All models were adjusted for propensity score and possible confounders, such as age, sex, and comorbidities. The analysis was performed separately for diagnostic and interventional catheterization.

RESULTS

The unadjusted relative risk (RR) of complications was not significantly different in diagnostic catheterization when a VCD was used (RR = 0.70; 95% confidence interval [CI] 0.22-2.16) but significantly lower in interventional catheterization (RR = 0.44; 95% CI 0.21-0.93). Costs were on average €275 lower in the diagnostic group (95% CI -€478.0 to -€64.9; P = 0.006) and around €373 lower in the interventional group (95% CI -€630.0 to -€104.2; P = 0.014) when a VCD was used. The adjusted estimated average length of stay did not differ significantly between the use of a VCD and manual compression in both types of catheterization.

CONCLUSIONS

In interventional catheterization, VCDs significantly reduced unadjusted complication rates, as well as costs. A significant reduction in costs also supports their usage in diagnostic catheterization on a larger scale.

Use of Systematic Methods to Improve Disease Identification in Administrative Data: The Case of Severe Sepsis

BACKGROUND

Selection of International Classification of Diseases (ICD)-based coded information for complex conditions such as severe sepsis is a subjective process and the results are sensitive to the codes selected. We use an innovative data exploration method to guide ICD-based case selection for severe sepsis.

METHODS

Using the Nationwide Inpatient Sample, we applied Latent Class Analysis (LCA) to determine if medical coders follow any uniform and sensible coding for observations with severe sepsis. We examined whether ICD-9 codes specific to sepsis (038.xx for septicemia, a subset of 995.9 codes representing Systemic Inflammatory Response syndrome, and 785.52 for septic shock) could all be members of the same latent class.

RESULTS

Hospitalizations coded with sepsis-specific codes could be assigned to a latent class of their own. This class constituted 22.8% of all potential sepsis observations. The probability of an observation with any sepsis-specific codes being assigned to the residual class was near 0. The chance of an observation in the residual class having a sepsis-specific code as the principal diagnosis was close to 0. Validity of sepsis class assignment is supported by empirical results, which indicated that in-hospital deaths in the sepsis-specific class were around 4 times as likely as that in the residual class.

CONCLUSIONS

The conventional methods of defining severe sepsis cases in observational data substantially misclassify sepsis cases. We suggest a methodology that helps reliable selection of ICD codes for conditions that require complex coding.

Upcoding in a National Health Service: the evidence from Portugal

For many years, evidence from the USA has pointed out to the existence of upcoding in management practices. Upcoding is defined as classifying patients in diagnosis-related groups codes associated with larger payments. The incentive for upcoding is not restricted to private providers of care. Conceptually, any patient classification system that is used for payment purposes may be vulnerable to this sort of strategic behaviour by providers. We document here that upcoding occurs in a National Health Service where public hospitals have their payment (budget) tied to the classification of treatment episodes. Using diagnosis-related groups data from Portugal, we found that the practice of upcoding has been used in the hospitals in a way leading to larger budgets (age of patients plays a key role). The effect is quantitatively small. Copyright © 2016 John Wiley & Sons, Ltd.

A modified Elixhauser score for predicting in-hospital mortality in internal medicine admissions

BACKGROUND

In-hospital mortality (IHM) is an indicator of the quality of care provided. The two most widely used scores for predicting IHM by International Classification of Diseases (ICD) codes are the Elixhauser (EI) and the Charlson Comorbidity indexes. Our aim was to obtain new measures based on internal medicine ICD codes for the original EI, to detect risk for IHM.

MATERIAL AND METHODS

This single-center retrospective study included hospital admissions for any cause in the department of internal medicine between January 1, 2000, and December 31, 2013, recorded in the hospital database. The EI was calculated for evaluation of comorbidity, then we added age, gender and diagnosis of ischemic heart disease. IHM was our outcome. Only predictors positively associated with IHM were taken into consideration and the Sullivan's method was applied in order to modify the parameter estimates of the regression model into an index.

RESULTS

We analyzed 75,586 admissions (53.4% females) and mean age was 72.7±16.3years. IHM was 7.9% and mean score was 12.1±7.6. The points assigned to each condition ranged from 0 to 16, and the possible range of the score varied between 0 and 89. In our population the score ranged from 0 to 54, and it was higher in the deceased group. Receiver operating characteristic curve of the new score was 0.721 (95% CI 0.714-0.727, p<0.001).

CONCLUSIONS

In order to make prognostic assessment, the use of a score could be of help in targeting interventions in older adults, identifying subjects at high risk for IHM.

Using Procedure Codes to Define Radiation Toxicity in Administrative Data: The Devil is in the Details

BACKGROUND

Systematic coding systems are used to define clinically meaningful outcomes when leveraging administrative claims data for research. How and when these codes are applied within a research study can have implications for the study validity and their specificity can vary significantly depending on treatment received.

SUBJECTS

Data are from the Surveillance, Epidemiology, and End Results-Medicare linked dataset.

STUDY DESIGN

We use propensity score methods in a retrospective cohort of prostate cancer patients first examined in a recently published radiation oncology comparative effectiveness study.

RESULTS

With the narrowly defined outcome definition, the toxicity event outcome rate ratio was 0.88 per 100 person-years (95% confidence interval, 0.71-1.08). With the broadly defined outcome, the rate ratio was comparable, with 0.89 per 100 person-years (95% confidence interval, 0.76-1.04), although individual event rates were doubled. Some evidence of surveillance bias was suggested by a higher rate of endoscopic procedures the first year of follow-up in patients who received proton therapy compared with those receiving intensity-modulated radiation treatment (11.15 vs. 8.90, respectively).

CONCLUSIONS

This study demonstrates the risk of introducing bias through subjective application of procedure codes. Careful consideration is required when using procedure codes to define outcomes in administrative data.