William A. Ghali

Coding algorithms for defining comorbidities in ICD-9-CM and ICD-10 administrative data.

Objectives:Implementation of the International Statistical Classification of Disease and Related Health Problems, 10th Revision (ICD-10) coding system presents challenges for using administrative data. Recognizing this, we conducted a multistep process to develop ICD-10 coding algorithms to define Charlson and Elixhauser comorbidities in administrative data and assess the performance of the resulting algorithms. Methods:ICD-10 coding algorithms were developed by “translation” of the ICD-9-CM codes constituting Deyos (for Charlson comorbidities) and Elixhausers coding algorithms and by physicians’ assessment of the face-validity of selected ICD-10 codes. The process of carefully developing ICD-10 algorithms also produced modified and enhanced ICD-9-CM coding algorithms for the Charlson and Elixhauser comorbidities. We then used data on in-patients aged 18 years and older in ICD-9-CM and ICD-10 administrative hospital discharge data from a Canadian health region to assess the comorbidity frequencies and mortality prediction achieved by the original ICD-9-CM algorithms, the enhanced ICD-9-CM algorithms, and the new ICD-10 coding algorithms. Results:Among 56,585 patients in the ICD-9-CM data and 58,805 patients in the ICD-10 data, frequencies of the 17 Charlson comorbidities and the 30 Elixhauser comorbidities remained generally similar across algorithms. The new ICD-10 and enhanced ICD-9-CM coding algorithms either matched or outperformed the original Deyo and Elixhauser ICD-9-CM coding algorithms in predicting in-hospital mortality. The C-statistic was 0.842 for Deyos ICD-9-CM coding algorithm, 0.860 for the ICD-10 coding algorithm, and 0.859 for the enhanced ICD-9-CM coding algorithm, 0.868 for the original Elixhauser ICD-9-CM coding algorithm, 0.870 for the ICD-10 coding algorithm and 0.878 for the enhanced ICD-9-CM coding algorithm. Conclusions:These newly developed ICD-10 and ICD-9-CM comorbidity coding algorithms produce similar estimates of comorbidity prevalence in administrative data, and may outperform existing ICD-9-CM coding algorithms.

Increasing incidence and prevalence of the inflammatory bowel diseases with time, based on systematic review.

BACKGROUND & AIMS We conducted a systematic review to determine changes in the worldwide incidence and prevalence of ulcerative colitis (UC) and Crohns disease (CD) in different regions and with time. METHODS We performed a systematic literature search of MEDLINE (1950-2010; 8103 citations) and EMBASE (1980-2010; 4975 citations) to identify studies that were population based, included data that could be used to calculate incidence and prevalence, and reported separate data on UC and/or CD in full manuscripts (n = 260). We evaluated data from 167 studies from Europe (1930-2008), 52 studies from Asia and the Middle East (1950-2008), and 27 studies from North America (1920-2004). Maps were used to present worldwide differences in the incidence and prevalence of inflammatory bowel diseases (IBDs); time trends were determined using joinpoint regression. RESULTS The highest annual incidence of UC was 24.3 per 100,000 person-years in Europe, 6.3 per 100,000 person-years in Asia and the Middle East, and 19.2 per 100,000 person-years in North America. The highest annual incidence of CD was 12.7 per 100,000 person-years in Europe, 5.0 person-years in Asia and the Middle East, and 20.2 per 100,000 person-years in North America. The highest reported prevalence values for IBD were in Europe (UC, 505 per 100,000 persons; CD, 322 per 100,000 persons) and North America (UC, 249 per 100,000 persons; CD, 319 per 100,000 persons). In time-trend analyses, 75% of CD studies and 60% of UC studies had an increasing incidence of statistical significance (P < .05). CONCLUSIONS Although there are few epidemiologic data from developing countries, the incidence and prevalence of IBD are increasing with time and in different regions around the world, indicating its emergence as a global disease.

The Canadian Adverse Events Study: the incidence of adverse events among hospital patients in Canada

Background: Research into adverse events (AEs) has highlighted the need to improve patient safety. AEs are unintended injuries or complications resulting in death, disability or prolonged hospital stay that arise from health care management. We estimated the incidence of AEs among patients in Canadian acute care hospitals. Methods: We randomly selected 1 teaching, 1 large community and 2 small community hospitals in each of 5 provinces (British Columbia, Alberta, Ontario, Quebec and Nova Scotia) and reviewed a random sample of charts for nonpsychiatric, nonobstetric adult patients in each hospital for the fiscal year 2000. Trained reviewers screened all eligible charts, and physicians reviewed the positively screened charts to identify AEs and determine their preventability. Results: At least 1 screening criterion was identified in 1527 (40.8%) of 3745 charts. The physician reviewers identified AEs in 255 of the charts. After adjustment for the sampling strategy, the AE rate was 7.5 per 100 hospital admissions (95% confidence interval [CI] 5.7– 9.3). Among the patients with AEs, events judged to be preventable occurred in 36.9% (95% CI 32.0%–41.8%) and death in 20.8% (95% CI 7.8%–33.8%). Physician reviewers estimated that 1521 additional hospital days were associated with AEs. Although men and women experienced equal rates of AEs, patients who had AEs were significantly older than those who did not (mean age [and standard deviation] 64.9 [16.7] v. 62.0 [18.4] years; p = 0.016). Interpretation: The overall incidence rate of AEs of 7.5% in our study suggests that, of the almost 2.5 million annual hospital admissions in Canada similar to the type studied, about 185 000 are associated with an AE and close to 70 000 of these are potentially preventable.

Association of alcohol consumption with selected cardiovascular disease outcomes: a systematic review and meta-analysis

Objective To conduct a comprehensive systematic review and meta-analysis of studies assessing the effect of alcohol consumption on multiple cardiovascular outcomes. Design Systematic review and meta-analysis. Data sources A search of Medline (1950 through September 2009) and Embase (1980 through September 2009) supplemented by manual searches of bibliographies and conference proceedings. Inclusion criteria Prospective cohort studies on the association between alcohol consumption and overall mortality from cardiovascular disease, incidence of and mortality from coronary heart disease, and incidence of and mortality from stroke. Studies reviewed Of 4235 studies reviewed for eligibility, quality, and data extraction, 84 were included in the final analysis. Results The pooled adjusted relative risks for alcohol drinkers relative to non-drinkers in random effects models for the outcomes of interest were 0.75 (95% confidence interval 0.70 to 0.80) for cardiovascular disease mortality (21 studies), 0.71 (0.66 to 0.77) for incident coronary heart disease (29 studies), 0.75 (0.68 to 0.81) for coronary heart disease mortality (31 studies), 0.98 (0.91 to 1.06) for incident stroke (17 studies), and 1.06 (0.91 to 1.23) for stroke mortality (10 studies). Dose-response analysis revealed that the lowest risk of coronary heart disease mortality occurred with 1–2 drinks a day, but for stroke mortality it occurred with ≤1 drink per day. Secondary analysis of mortality from all causes showed lower risk for drinkers compared with non-drinkers (relative risk 0.87 (0.83 to 0.92)). Conclusions Light to moderate alcohol consumption is associated with a reduced risk of multiple cardiovascular outcomes.

D-Dimer for the Exclusion of Acute Venous Thrombosis and Pulmonary Embolism: A Systematic Review

Context Clinicians may not know which d-dimer assay is best for diagnosing deep venous thrombosis (DVT) or pulmonary embolism (PE). Contribution This meta-analysis summarizes data from 78 prospective studies that compared results of different d-dimer assays with findings of objective tests (for example, compression ultrasonography, venography, lung scanning) in patients with suspected DVT or PE. Enzyme-linked immunosorbent assays (ELISAs) had the best sensitivity (about 95%) and negative likelihood ratios (about 0.1) for excluding DVT and PE. None of the assays had positive likelihood values that greatly increased the certainty of diagnosis. Implications Negative ELISA results are strong evidence against DVT or PE. The Editors Tests for d-dimer to exclude venous thromboembolic disease have been available since the 1980s (1). Hundreds of original studies, reviews, commentaries, and technical notes related to d-dimer as a diagnostic aid have been published. Even with this extensive literature, the role of d-dimer in the diagnosis of deep venous thrombosis (DVT) or pulmonary embolism (PE) remains unclear. This lack of clarity is due in part to the multiple d-dimer assays that are available (Appendix Table 1), availability of central laboratory and point-of-care testing, and concerns about differing sensitivities and variability of the assays (1). In 1998, the American College of Chest Physicians Consensus Panel on Pulmonary Embolism called for more evaluation of the utility of d-dimer (2). In 1999, a Clinical Practice Guideline of the American Thoracic Society also called for additional outcome studies to further define the role of rapid bedside assays of d-dimer (3). As recently as 2000, investigators argued that the whole-blood agglutination test with d-dimer is not validated for clinical use for the exclusion of DVT (4). Although several useful reviews of the topic exist (1, 3, 5), none in recent years have comprehensively reviewed the world literature in an effort to summarize the accumulated published experience with d-dimer testing. Given continuing uncertainty about d-dimer testing, we performed a systematic review of the literature to assess the sensitivity and specificity of the d-dimer assays and the variability of those measures among studies for diagnosing DVT and PE. The comparative findings among the differing d-dimer assays provide both the laboratory pathologist and clinician with a practical pathway for translating clinical research into practice. Methods We used several sources to guide our review processes, including recommendations by Lijmer and colleagues concerning avoidance of bias in studies of diagnostic tests (6), the Standards for Reporting Diagnostic Accuracy (STARD) statement (7, 8), and guidelines for meta-analyses of observational studies in epidemiology (9). Study Identification We attempted to identify all published trials in all languages that used d-dimer to exclude PE or DVT on the basis of objective diagnostic tests. Studies were identified by searching PubMed from 1983 to January 2003 and EMBASE from 1988 to January 2003. All searches used the key words d -dimer, PE, and DVT. We augmented our searches by manually reviewing the reference lists of all original articles and all review articles. This was done by 2 of the authors working together. Abstracts were excluded. Study Eligibility At least 2 authors evaluated each study for inclusion. Any disagreements were resolved by discussion. Authors were not blinded to journal, author, or institution. Studies were included if they met all of the following criteria: 1) A specific statement was made about whether PE or DVT (not the inclusive term thromboembolic disease) was being diagnosed; 2) the diagnosis of PE or DVT was based on objective tests; 3) studies were performed prospectively; 4) participants were recruited consecutively; 5) the population studied included a broad spectrum of patients; 6) results of d-dimer and the diagnostic tests for PE and DVT were interpreted independently; 7) the participants studied were suspected of having PE or DVT, and all studies included patients with and without disease; 8) the decision to perform the reference diagnostic test was made independently of the d-dimer result; 9) test descriptions were sufficiently detailed to permit replication; 10) the cutoff value for a negative d-dimer test result was stated unless qualitative tests were used; and 11) sensitivity and specificity or the raw data for these calculations were presented. We labeled studies that did not meet the third, fourth, or tenth inclusion criteria as tier 3 studies and included them in sensitivity analyses. We categorized studies that met all inclusion criteria into 2 tiers. Tier 1 included studies that compared an enzyme-linked immunosorbent assay (ELISA) and at least one other d-dimer assay. Tier 2 included the tier 1 studies and all other studies that met all inclusion criteria. Data Extraction Two authors collected data on the following study-level factors: 1) the d-dimer assay used in the study, 2) the cutoff value below which disease was considered to be absent, and 3) whether d-dimer was used to exclude PE or DVT. At least 2 authors confirmed the values for sensitivity and specificity. Statistical Analysis Sensitivities reflect the proportion of patients with disease who had a positive d-dimer result, while specificities reflect the proportion of patients without disease who had a negative d-dimer result, depending on cutoff level. Values for the likelihood ratio (the multiplicative factor for converting pretest to post-test disease probabilities) associated with positive test results were obtained by the following formula: sensitivity/(1 corresponding specificity). This formula provided likelihood ratios arising from negative test results. Primary analysis was restricted to the 500-ng/mL cutoff because that was the most commonly used value. On the basis of the varying array of assays examined in individual studies and our anticipation of important between-study heterogeneity, we applied a linear mixed-model approach (10) to jointly analyze the proportions of positive test results in the disease and nondisease samples (that is, true-positive and false-positive rates). Similar models have been applied to Bayesian meta-analysis of receiver-operating characteristic curves (11). The main explanatory term in the model was a fixed effect reflecting distinct positive response rates for each combination of assay and population. To allow for variability in assay performance that might arise from idiosyncratic features of patient samples, such as spectrum of disease severity, and other aspects of study context, such as laboratory procedure, we incorporated 3 random-effects terms corresponding to assay nested within patient group (disease/nondisease) nested within study. Residual variances were assumed to follow the standard binomial form depending on underlying proportion and sample size. Applying restricted maximum likelihood, we obtained population average estimates for sensitivity and specificity, which were later combined to provided estimated likelihood ratios. Overall, the estimates did not differ substantially from conventional sample-size weighted averages, although the associated standard errors were increased, reflecting underlying between-study heterogeneity. The joint statistical significance of overall differences was assessed by using likelihood-based Wald tests. Pairwise differences between estimates were assessed on the basis of the model-based standard errors without adjustment for multiple comparisons. Significant pairwise differences should be interpreted with caution when the Wald test is not significant. Confidence limits are derived from asymptotic standard errors; asymmetric intervals for likelihood ratios reflect the application of likelihood theory to logarithmically transformed estimates. Values for the sensitivity and specificity for the different studies and test types were examined graphically by use of boxplots. The range between the upper and lower quartiles of the values for each assay provides a measure of between-study variability associated with the assay. Sensitivity analyses were conducted by adding the 30 studies that did not meet one or more of the 3 inclusion criteria previously mentioned (tier 3 analysis) and by analyzing sensitivity and specificity at the 250-ng/mL and 1000-ng/mL cutoffs. All analyses were conducted by using S-PLUS, version 6.1.2, 2002 (Insightful Corp., Seattle, Washington) (12). Data Synthesis Figure 1 summarizes our search. We initially identified 513 potentially relevant reports. Thirty-one studies (13-43) were included; they met a priori inclusion criteria and compared 2 or more assays, including an ELISA (tier 1 analysis). An additional 47 studies (44-90) were also included; they did not have a comparative ELISA but met the a priori inclusion criteria evaluating a d-dimer assay. The 47 studies combined with the 31 studies provided a study sample of 78 studies (tier 2 analysis). Thirty additional studies (91-120) that did not meet one or more inclusion criteria were included in the sensitivity, or tier 3, analysis, which also included the 78 studies in the tier 2 analysis. Figure 1. Reports evaluated for inclusion in the review. Appendix Tables 2 and 3 list the d-dimer assays that were evaluated, patient characteristics, and the objective diagnostic reference test that was used for patients with clinically suspected DVT (49 studies) and clinically suspected PE (31 studies). All studies met the 11 criteria for inclusion listed in the Methods section. There was a total of 78 studies (2 studies [60, 71] gave data for patients with PE and DVT separately in the same article). In the various studies, prevalence of DVT ranged from 20% to 78% (overall prevalence, 36%), and the prevalence of PE ranged from 8% to 62% (overall prevalence, 25%). Sixteen studies of DVT used comp

Physician wellness: a missing quality indicator

When physicians are unwell, the performance of health-care systems can be suboptimum. Physician wellness might not only benefit the individual physician, it could also be vital to the delivery of high-quality health care. We review the work stresses faced by physicians, the barriers to attending to wellness, and the consequences of unwell physicians to the individual and to health-care systems. We show that health systems should routinely measure physician wellness, and discuss the challenges associated with implementation.

Effect of alcohol consumption on biological markers associated with risk of coronary heart disease: systematic review and meta-analysis of interventional studies

Objective To systematically review interventional studies of the effects of alcohol consumption on 21 biological markers associated with risk of coronary heart disease in adults without known cardiovascular disease. Design Systematic review and meta-analysis. Data sources Medline (1950 to October 2009) and Embase (1980 to October 2009) without limits. Study selection Two reviewers independently selected studies that examined adults without known cardiovascular disease and that compared fasting levels of specific biological markers associated with coronary heart disease after alcohol use with those after a period of no alcohol use (controls). 4690 articles were screened for eligibility, the full texts of 124 studies reviewed, and 63 relevant articles selected. Results Of 63 eligible studies, 44 on 13 biomarkers were meta-analysed in fixed or random effects models. Quality was assessed by sensitivity analysis of studies grouped by design. Analyses were stratified by type of beverage (wine, beer, spirits). Alcohol significantly increased levels of high density lipoprotein cholesterol (pooled mean difference 0.094 mmol/L, 95% confidence interval 0.064 to 0.123), apolipoprotein A1 (0.101 g/L, 0.073 to 0.129), and adiponectin (0.56 mg/L, 0.39 to 0.72). Alcohol showed a dose-response relation with high density lipoprotein cholesterol (test for trend P=0.013). Alcohol decreased fibrinogen levels (−0.20 g/L, −0.29 to −0.11) but did not affect triglyceride levels. Results were similar for crossover and before and after studies, and across beverage types. Conclusions Favourable changes in several cardiovascular biomarkers (higher levels of high density lipoprotein cholesterol and adiponectin and lower levels of fibrinogen) provide indirect pathophysiological support for a protective effect of moderate alcohol use on coronary heart disease.

Comparison of the Elixhauser and Charlson/Deyo methods of comorbidity measurement in administrative data.

Background:Comorbidity risk adjustment methods have been used widely with administrative data, and the Charlson/Deyo method is perhaps the most commonly used in the literature. However, a new method defined by Elixhauser et al. has been introduced recently and could be superior, although it has not been validated widely. Objectives:We compared the Charlson/Deyo and Elixhauser methods using Canadian administrative data on patients with myocardial infarction (MI). Research Design:We conducted a historical cohort study. Subjects:We used administrative hospital discharge data from a large Canadian city for all cases with acute MI coded as most responsible diagnosis between January 1, 1995, and March 31, 2001. Measures:We used each of the 2 methods to define comorbidity variables based on the International Classification of Diseases, 9th Revision, Clinical Modification codes present in each case record. We then compared 2 models predicting in-hospital mortality based on presence or absence of the variables defined by each of the methods. Frequency tables were produced and c-statistics and changes in −2 log likelihood (−2LogL) were calculated. We also visually assessed model performance by plotting observed and expected percentages of death for increasing risk categories defined by the 2 models. Results:The Elixhauser model outperformed the Charlson/Deyo model in predicting mortality, with higher c-statistic values (0.793 vs. 0.704). Superior performance of the Elixhauser method is confirmed when plotting the expected and observed risks of death across groupings of increasing risk, in which the Elixhauser method yields a wider range of predicted and observed probabilities of death across groupings (2.5%–33%) than does the Charlson/Deyo method (5%–25%). Conclusions:The Elixhauser comorbidity measurement method performs better than the widely used Charlson/Deyo method in the Canadian acute MI cases studied.

A comprehensive view of sex-specific issues related to cardiovascular disease

Cardiovascular disease (CVD) is the leading cause of mortality in women. In fact, CVD is responsible for a third of all deaths of women worldwide and half of all deaths of women over 50 years of age in developing countries. The prevalence of CVD risk factor precursors is increasing in children. Retrospective analyses suggest that there are some clinically relevant differences between women and men in terms of prevalence, presentation, management and outcomes of the disease, but little is known about why CVD affects women and men differently. For instance, women with diabetes have a significantly higher CVD mortality rate than men with diabetes. Similarly, women with atrial fibrillation are at greater risk of stroke than men with atrial fibrillation. Historically, women have been underrepresented in clinical trials. The lack of good trial evidence concerning sex-specific outcomes has led to assumptions about CVD treatment in women, which in turn may have resulted in inadequate diagnoses and suboptimal management, greatly affecting outcomes. This knowledge gap may also explain why cardiovascular health in women is not improving as fast as that of men. Over the last decades, mortality rates in men have steadily declined, while those in women remained stable. It is also becoming increasingly evident that gender differences in cultural, behavioural, psychosocial and socioeconomic status are responsible, to various degrees, for the observed differences between women and men. However, the interaction between sex-and gender-related factors and CVD outcomes in women remains largely unknown.

Validity of procedure codes in International Classification of Diseases, 9th revision, clinical modification administrative data.

Background:Administrative hospital discharge data are widely used to assess quality of care in patients undergoing certain procedures. However, little is known about the validity of administrative coding of procedure data. We conducted a detailed chart review to evaluate the accuracy and completeness of information on procedures in administrative data. Methods:We randomly selected 1200 hospital separations in the period April 1, 1996, to March 31, 1997, from administrative discharge data of 3 acute adult hospitals in Calgary, Alberta, Canada. Each separation record in administrative data contains up to 10 procedure coding fields. The corresponding medical charts were reviewed for recording presence or absence of procedures. We then determined sensitivity to quantify the accuracy of coding presence of procedures in administrative data when these are present in the chart data (criterion standard). Results:The agreement between the 2 databases varied greatly across 35 procedures studied. The sensitivity ranged from 0% to 94%. Of 6 major procedures studied, validity of coding was generally good, with 5 procedures having coding sensitivity of 69% and over and only 1 (lysis of peritoneal adhesion) with a low sensitivity of 41%. In contrast, many minor procedures had low sensitivities. Of 29 minor procedures studied, sensitivity was lower than 50% for 15 procedures, between 50% and 79% for 10, and 80% and over for 4. Conclusion:Validity of information on procedures in administrative discharge data appears to be related to type of procedures. Major procedures that are usually performed in operating rooms are reasonably well-coded. Meanwhile, minor procedures that are routinely performed on wards or in radiology departments are generally undercoded.