Administrative Data for Quality Improvement

BackgroundBecause the collection of mental health information through interviews is expensive and time consuming, interest in using population-based administrative health data to conduct research on depression has increased. However, there is concern that misclassification of disease diagnosis in the underlying data might bias the results. Our objective was to determine the validity of International Classification of Disease (ICD)-9 and ICD-10 administrative health data case definitions for depression using review of family physician (FP) charts as the reference standard.MethodsTrained chart reviewers reviewed 3362 randomly selected charts from years 2001 and 2004 at 64 FP clinics in Alberta (AB) and British Columbia (BC), Canada. Depression was defined as presence of either: 1) documentation of major depressive episode, or 2) documentation of specific antidepressant medication prescription plus recorded depressed mood. The charts were linked to administrative data (hospital discharge abstracts and physician claims data) using personal health numbers. Validity indices were estimated for six administrative data definitions of depression using three years of administrative data.ResultsDepression prevalence by chart review was 15.9–19.2% depending on year, region, and province. An ICD administrative data definition of ‘2 depression claims with depression ICD codes within a one-year window OR 1 discharge abstract data (DAD) depression diagnosis’ had the highest overall validity, with estimates being 61.4% for sensitivity, 94.3% for specificity, 69.7% for positive predictive value, and 92.0% for negative predictive value. Stratification of the validity parameters for this case definition showed that sensitivity was fairly consistent across groups, however the positive predictive value was significantly higher in 2004 data compared to 2001 data (78.8 and 59.6%, respectively), and in AB data compared to BC data (79.8 and 61.7%, respectively).ConclusionsSensitivity of the case definition is often moderate, and specificity is often high, possibly due to undercoding of depression. Limitations to this study include the use of FP charts data as the reference standard, given the potential for missed or incorrect depression diagnoses. These results suggest that that administrative data can be used as a source of information for both research and surveillance purposes, while remaining aware of these limitations.

BackgroundSyndromic surveillance, which provides real-time data, may provide timely drowning surveillance compared with hospital discharge data where the release of data may be delayed. We compared data on hospital visits...

BackgroundInforming health systems and monitoring hospital performances using administrative data sets, mainly hospital discharge data coded according to International-Classification-Diseases-9edition-Clinical-Modifiers (ICD9-CM), is now commonplace in several countries, but the reliability of diagnostic coding of acute ischemic stroke in the routine practice is uncertain. This study aimed at estimating accuracy of ICD9-CM codes for the identification of acute ischemic stroke and the use of thrombolysis treatment comparing hospital discharge data with medical record review in all the six hospitals of the Florence Area, Italy, through 2015.MethodsWe reviewed the medical records of all the 3915 potential acute stroke events during 2015 across the six hospitals of the Florence Area, Italy. We then estimated sensitivity and Positive Predictive Value of ICD9-CM code-groups 433*1, 434*1 and thrombolysis code 99.10 against medical record review with clinical adjudication. For each false-positive case we obtained the actual diagnosis. For each false-negative case we obtained the primary and secondary ICD9-CM diagnoses.ResultsThe medical record review identified 1273 acute ischemic stroke events. The hospital discharge records identified 898 among those (true-positive cases),but missed 375 events (false-negative cases), and identified 104 events that were not eventually confirmed as acute ischemic events (false-positive cases). Code-group specific Positive Predictive Value was 85.7% (95%CI,74.6–93.3) for 433*1 and 89.9% (95%CI, 87.8–91.7) for 434*1 codes. Thrombolysis treatment, as identified by ICD9-CM code 99.10, was only documented in 6.0% of acute ischemic stroke events, but was 13.6% in medical record review.ConclusionsHospital discharge data were found to be fairly specific but insensitive in the reporting of acute ischemic stroke and thrombolysis, providing misleading indications about both quantity and quality of acute ischemic stroke hospital care. Efforts to improve coding accuracy should precede the use of hospital discharge data to measure hospital performances in acute ischemic stroke care.

The objective of our study was to examine the prevalence of diabetes during pregnancy at the population level in SC from January 1996 through December 2008. The study included 387,720 non-Hispanic white (NHW), 232,278 non-Hispanic black (NHB), and 43,454 Hispanic live singleton births. Maternal inpatient hospital discharge codes from delivery (91.59%) and prenatal information (i.e., Medicaid [42.91%] and SC State Health Plan [SHP] [5.98%]) were linked to birth certificate data. Diabetes during pregnancy included gestational and preexisting, defined by prenatal and maternal inpatient International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) diagnostic codes (i.e., 64801-64802, 64881-64882, or 25000-25092) or report on the birth certificate. Diabetes prevalence from any source increased from 5.02% (95% confidence interval [CI]: 4.82-5.22) in 1996 to 8.37% (95% CI: 8.15-8.60) in 2008. Diabetes prevalence, standardized for maternal age and race/ethnicity from 1996 through 2008, increased from 3.38% (95% CI: 3.29-3.47) to 5.81% (95% CI: 5.71-5.91) using birth certificate data, from 3.99% (95% CI: 3.89-4.10) to 6.69% (95% CI: 6.58-6.80) using hospital discharge data, and from 4.74% (95% CI: 4.52-4.96) to 8.82% (95% CI: 8.61-9.03) using Medicaid data. Comparing birth certificate to hospital discharge, Medicaid, and SHP data, Cohen's kappa in 2008 was 0.73 (95% CI: 0.72-0.75), 0.64 (95% CI: 0.62-0.66), and 0.59 (95% CI: 0.54-0.65), respectively. An increasing prevalence of diabetes during pregnancy is reported, as well as substantial lack of agreement in reporting of diabetes prevalence across administrative databases. Prevalence of reported diabetes during pregnancy is impacted by screening, diagnostic, and reporting practices across different data sources, as well as by actual changes in prevalence over time.

Validation of an Electronic Surveillance Tool for Identifying Hospital Inpatients With Severe Sepsis

This work summarizes how hospital discharge data are used, identifies strengths and shortcomings, and presents suggestions for enhancing usefulness of the data. Results demonstrate that discharge data are used in a wide range of applications by diverse users. Uses include public health and population-based applications, as well as quality assessment, informed purchasing, strategic planning, and policy making. Strategies to enhance the utility of discharge data include: improving the quality of existing data elements and adding new data elements that will support more advanced analyses, improving linkages with data from nonhospital settings and databases outside health care, and developing a technical assistance network to support statewide data organizations in their efforts to collect and analyze discharge data. As our nation moves toward universal electronic medical records, it will be important to keep in mind the many uses of discharge data in order to maintain the data capacity to fill these needs.

This study aimed to examine the validity of using Maryland hospital discharge data to characterize injuries sustained by trauma patients. Maryland hospital discharge and Maryland trauma registry data for 1999 were merged, and the extent of agreement regarding the presence and severity of injuries sustained was evaluated. The mean Injury Severity Score was 8.4 according to the Maryland hospital discharge data and 10 according to the Maryland trauma registry data (p < 0.0001). The Maryland hospital discharge data identified 95% or more of all moderate to severe injuries (Abbreviated Injury Score, > or =2) for all body regions except the head. There was substantial agreement between the two data sets for mechanism of injury (weighted kappa, 0.62), the number of preexisting conditions present (weighted kappa, 0.45) and final disposition (weighted kappa, 0.78). The Maryland hospital discharge data are a valid source for documenting the nature and severity of injuries sustained by trauma patients, except for those with a relatively minor head injury.

Background and PurposeThe claims data of the Korean National Health Insurance (NHI) system can be useful in stroke research. The aim of this study was to validate the accuracy of hospital discharge data used for NHI claims in identifying acute stroke and use of thrombolytic therapy.MethodsThe hospital discharge data of 1,811 patients with stroke-related diagnosis codes were obtained from Jeju National University Hospital (JNUH) and Seoul Medical Center (SMC). Three algorithms were tested to identify discharges with acute stroke [ischemic stroke (IS), intracranial hemorrhage (ICH), or subarachnoid hemorrhage (SAH)]: 1) all diagnosis codes up to nine positions, 2) one primary diagnosis and one secondary diagnosis, and 3) only one primary diagnosis code. Reviews of medical records were considered the gold standards.ResultsOverall, the degree of agreement (κ) was higher for algorithms 1 and 2 than for algorithm 3, and the sensitivity and specificity of the first two algorithms for IS and SAH were both >90%, with almost perfect agreement (κ=0.83-0.84) in the JNUH data set. Regarding ICH, only algorithm 1 yielded an almost perfect agreement (κ=0.82). In the SMC data set, almost perfect agreement was found for both ICH and SAH in all three algorithms. In contrast, the three algorithms yielded a range of agreement levels, though all substantial, for IS. Almost perfect agreement was obtained for use of thrombolytic therapy in both data sets (κ=0.91-0.99).ConclusionsDischarge with hemorrhagic stroke and use of thrombolytic therapy were identified with high reliability in administrative discharge data. A substantial level of agreement was also obtained for IS, despite variation between the algorithms and data sets.

To compare quality outcomes from surgical procedures performed at freestanding ambulatory surgery centers (ASCs) and hospital-based outpatient departments (HOPDs). Patient-level ambulatory surgery (1997-2004), hospital discharge (1997-2004), and vital statistics data (1997-2004) for the state of Florida were assembled and analyzed. We used a pooled, cross-sectional design. Logistic regressions with time fixed-effects were estimated separately for the 12 most common ambulatory surgical procedures. Our quality outcomes were risk-adjusted 7-day and 30-day mortality and 7-day and 30-day unexpected hospitalizations. Risk-adjustment for patient demographic characteristics and severity of illness were calculated using the DCG/HCC methodology adjusting for primary diagnosis only and separately for all available diagnoses. Although neither ASCs nor HOPDs performed better overall, we found some difference by procedure that varied based on the risk-adjustment approach used. There appear to be important variations in quality outcomes for certain procedures, which may be related to differences in organizational structure, processes, and strategies between ASCs and HOPDs. The study also confirms the importance of risk-adjustment for comorbidities when using administrative data, particularly for procedures that are sensitive to differences in severity.

There are no nursing centric data in the hospital discharge abstract. This study investigates whether adding nursing data in the form of nursing diagnoses to medical diagnostic data in the discharge abstract can improve overall explanation of variance in commonly studied hospital outcomes. A retrospective analyses of 123,241 sequential patient admissions to a university hospital in a Midwestern city was performed. Two data sets were combined: (1) a daily collection of patient assessments by nurses using nursing diagnosis terminology (NDX); and (2) the summary discharge information from the hospital discharge abstract including diagnosis-related group (DRG) and all payer refined DRG (APR-DRG). Each of 61 daily NDX observations were collapsed as frequency of occurrence for the hospital stay and inserted into the discharge abstract. NDX was then compared to both DRG and APR-DRG across 5 hospital outcome variables using multivariate regression or logistic regression. In all statistical models, DRG, APR-DRG, and NDX were significantly associated with the 5 hospital outcome variables (P <.0001). When NDX was added to models containing either the DRG or the APR-DRG, explanatory power (R2) and model discrimination (c statistic) improved by 30% to 146% across the outcome variables of hospital length of stay, ICU length of stay, total charges, probably of death, and discharge to a nursing home (P <.0001). The findings support the contention that nursing care is an independent predictor of patient hospital outcomes. These nursing data are not redundant with the medical diagnosis, in particular, the DRG. The findings support the argument for including nursing care data in the hospital discharge abstract. Further study is needed to clarify which nursing data are the best fit for the current hospital discharge abstract data collection scheme.

Hospital discharge data can be used for monitoring procedures and intensive care related to severe maternal morbidity

Validity of Birth Certificate and Hospital Discharge Data Reporting of Labor Induction

Emergency room and hospital discharge data have been used to describe the risk factors and public health impact of dog bites. These data sets are based on financial charges for severe bites and underestimates dog bite burdens within communities. This study expands both the source of information and risk factor data collected to provide demographic analysis of dog bite injury risk factors reported in Bay County, Florida in 2009-2010. Extended data for dog bites reported by various sources from January 1, 2009 to December 31, 2010 were collected by Florida Department of Health in Bay County. Data collected included bite victim's age and gender, primary reported cause of bite, setting, dog's restraint status and relationship between the victim and the dog. A total of 799 bites were reported. Most bites (55%) were reported first by healthcare practitioners, particularly bites involving children<6 years. Bites involving unfamiliar dogs and dogs off the owner's property were more likely to be reported by other sources. Boys aged 6-14 years accounted for 2.24 times more bites than same-aged females (P<0.001) and had the highest incidence with 424 bites per 100,000 persons per year. Persons 6 years or older were 3.6 times more likely to be bitten by an unfamiliar dog. Inappropriate behaviour management was the most common cause of bites (26%), followed by protective behaviour (24%). Bites of unknown cause were 2.5 times more likely in children<6 years. Separating dog fights was the most common cause of bites for persons 15 years or older (24%); females were significantly more likely to be bit than males (P=0.01). Bites by unrestrained dogs off the owner's property (32% of all bites) most commonly involved males. Estimates based solely on healthcare discharge data significantly underestimate dog bite burden within a community. Characterizing these risks by age group or gender provides an opportunity to implement targeted interventions to prevent dog bites.

Introduction: Race is known to affect the quality and delivery of care for multiple disease states. The incidence of inflammatory bowel disease (IBD) is rising among racial minorities, and healthcare disparities are becoming increasingly obvious. We therefore sought to identify the effect of race or ethnicity on IBD hospitalization outcomes. Methods: Inpatient discharge data from a database covering more than 90% of nonprofit academic hospitals in the US and their affiliated community centers (the Vizient clinical database) was used to answer this question. Hospital discharge data from 10/1/2015 to 3/31/2018 was critically analyzed. Hospital discharges were identified by a principal ICD-10 diagnosis of ulcerative colitis (UC) or Crohn's disease (CD). Patients who were transferred to another hospital were excluded. Unadjusted outcomes between racial (White, Black, or Asian) or ethnic (Hispanic) groups were compared using the one-way ANOVA and Pearson chi-squared test. Using multivariate linear regression of aggregated hospital discharge data, we examined the effect of race or ethnicity on mean length of stay (LOS), mean direct cost, and 30-day readmissions while adjusting for potential confounders. Results: We identified a total of 19,063 hospitalizations for UC and 37,838 hospitalizations for CD among 323 medical centers (152 academic, 184 community). Unadjusted hospitalization outcomes by race or ethnicity are presented in table 1. For UC, black patients had a shorter mean LOS (5.68 days, p<0.01) compared to other racial or ethnic groups. For CD, Hispanic patients had a lower mean direct cost ($6,415, p<0.01) and black patients had a higher 30-day readmission rate (12.0%, p<0.01) compared to other racial or ethnic groups. After multivariate linear regression to control for confounders, the only significant association was lower mean direct cost for black and white patients in UC (table 2) and CD (table 3) hospitalizations. Conclusion: Racial and ethnic disparities were seen only for unadjusted outcomes among hospitalized IBD patients. When controlling for other demographics, hospital factors, disease severity, comorbidities, and payer status, these effects were not appreciated. This suggests that there is no observable difference for mean LOS, mean direct hospitalization cost, and readmission rate that is attributable to race or ethnic background. Future prospective evaluation is appropriate to validate these findings.643_A Figure 1. *P-values for compared means and frequencies were calculated using one-way ANOVA and chi-squared tests, respectively. LOS = length of stay, SD = standard deviation.643_B Figure 2. LOS = length of stay, CMI = case mix index, OSH = outside hospital, SOI = severity of illness, ICU = intensive care unit. *Reference point for White, Black, and Asian race and for Hispanic ethnicity was “Other or unreported.”643_C Figure 3. LOS = length of stay, CMI = case mix index, OSH = outside hospital, SOI = severity of illness, ICU = intensive care unit. *Reference point for White, Black, and Asian race and for Hispanic ethnicity was “Other or unreported.”

Identification of ICD codes suggestive of child maltreatment