Breast cancer incidence using administrative data: correction with sensitivity and specificity

Adjusting Incidence Estimates with Laboratory Test Performances: A Pragmatic Maximum Likelihood Estimation-Based Approach

Understanding the incidence of disease is often crucial for public policy decision-making, as observed during the COVID-19 pandemic. Estimating incidence is challenging, however, when the definition of incidence relies on tests that imperfectly measure disease, as in the case when assays with variable performance are used to detect the SARS-CoV-2 virus. To our knowledge, there are no pragmatic methods to address the bias introduced by the performance of labs in testing for the virus. In the setting of a longitudinal study, we developed a maximum likelihood estimation-based approach to estimate laboratory performance-adjusted incidence using the expectation-maximization algorithm. We constructed confidence intervals (CIs) using both bootstrapped-based and large-sample interval estimator approaches. We evaluated our methods through extensive simulation and applied them to a real-world study (TrackCOVID), where the primary goal was to determine the incidence of and risk factors for SARS-CoV-2 infection in the San Francisco Bay Area from July 2020 to March 2021. Our simulations demonstrated that our method converged rapidly with accurate estimates under a variety of scenarios. Bootstrapped-based CIs were comparable to the large-sample estimator CIs with a reasonable number of incident cases, shown via a simulation scenario based on the real TrackCOVID study. In more extreme simulated scenarios, the coverage of large-sample interval estimation outperformed the bootstrapped-based approach. Results from the application to the TrackCOVID study suggested that assuming perfect laboratory test performance can lead to an inaccurate inference of the incidence. Our flexible, pragmatic method can be extended to a variety of disease and study settings.

Validation of an algorithm for identifying incident cancer cases based on long-term illness and diagnosis related group program data from the French National Health Insurance Information System (SNDS)

PURPOSE

Three generic claims-based algorithms based on the Illness Classification of Diseases (10th revision- ICD-10) codes, French Long-Term Illness (LTI) data, and the Diagnosis Related Group program (DRG) were developed to identify retirees with cancer using data from the French national health insurance information system (Système national des données de santé or SNDS) which covers the entire French population. The present study aimed to calculate the algorithms' performances and to describe false positives and negatives in detail.

METHODS

Between 2011 and 2016, data from 7544 participants of the French retired self-employed craftsperson cohort (ESPrI) were first matched to the SNDS data, and then toFrench population-based cancer registries data, used as the gold standard. Performance indicators, such as sensitivity and positive predictive values, were estimated for the three algorithms in a subcohort of ESPrI.

RESULTS

The third algorithm, which combined the LTI and DRG program data, presented the best sensitivities (90.9%-100%) and positive predictive values (58.1%-95.2%) according to cancer sites. The majority of false positives were in fact nearby organ sites (e.g., stomach for esophagus) and carcinoma in situ. Most false negatives were probably due to under declaration of LTI.

CONCLUSION

Validated algorithms using data from the SNDS can be used for passive epidemiological follow-up for some cancer sites in the ESPrI cohort.

Prediction of early death after atrial fibrillation diagnosis using a machine learning approach: A French nationwide cohort study

AIMS

Atrial fibrillation is associated with important mortality but the usual clinical risk factor based scores only modestly predict mortality. This study aimed to develop machine learning models for the prediction of death occurrence within the year following atrial fibrillation diagnosis and compare predictive ability against usual clinical risk scores.

METHODS AND RESULTS

We used a nationwide cohort of 2,435,541 newly diagnosed atrial fibrillation patients seen in French hospitals from 2011 to 2019. Three machine learning models were trained to predict mortality within the first year using a training set (70% of the cohort). The best model was selected to be evaluated and compared with previously published scores on the validation set (30% of the cohort). Discrimination of the best model was evaluated using the C index. Within the first year following atrial fibrillation diagnosis, 342,005 patients (14.4%) died after a period of 83 (SD 98) days (median 37 [10-129]). The best machine learning model selected was a deep neural network with a C index of 0.785 (95% CI, 0.781-0.789) on the validation set. Compared to clinical risk scores, the selected model was superior to the CHA2DS2-VASc and HAS-BLED risk scores and superior to dedicated scores such as Charlson Comorbidity Index and Hospital Frailty Risk Score to predict death within the year following atrial fibrillation diagnosis (C indexes: 0.597; 0.562; 0.643; 0.626 respectively. P < .0001).

CONCLUSION

Machine learning algorithms predict early death after atrial fibrillation diagnosis and may help clinicians to better risk stratify atrial fibrillation patients at high risk of mortality.

Validity of claims data for identifying cancer incidence in the Japan public health center-based prospective study for the next generation

PURPOSE

This study determined the validity of claims-based definitions for identifying the incidence of total and site-specific cancers in a population-based cohort study.

METHODS

Claims data were obtained for 21 946 participants aged 40-74 years enrolled in the Japan Public Health Center-based Prospective Study for the Next Generation. We defined total and site-specific cancer incidence using combinations of codes from claims data, including diagnosis and procedure codes for cancer therapy. Data from the cancer registry were used as the gold standard to evaluate validity.

RESULTS

Among 21 946 participants, 454 total, 89 stomach, 67 colorectal, 51 lung, 39 breast and 99 prostate invasive cancer cases were newly diagnosed in the cancer registry. For invasive cancer, the sensitivity and specificity of the definition that combined codes for diagnosis and procedures for cancer therapy were 87.0% and 99.4% for total, 88.8% and 99.9% for stomach, 80.6% and 99.9% for colorectal, 86.3% and 99.9% for lung, 100% and 99.9% for breast and 91.9% and 99.9% for prostate cancer, respectively. Furthermore, for invasive and/or in situ cancer, the sensitivity and specificity of the definition were 84.5% and 99.5% for total, 66.7% and 99.9% for colorectal and 100% and 99.9% for breast cancer.

CONCLUSIONS

Our findings suggest that claims-based definitions using diagnosis and procedure codes generally have high validity for total, stomach, lung, breast and prostate cancer incidence, but may underestimate colorectal cancer incidence.

Epidemiology of interstitial lung disease in systemic lupus erythematosus in France: A nation-wide population-based study over 10 years

Background and objective

Data regarding interstitial lung disease (ILD) in the setting of systemic lupus erythematosus (SLE) are limited. We used a nationwide database to determine the incidence and the prevalence of ILD in SLE.

Methods

Characteristics of all SLE inpatients admitted between 2011 and 2012 in France were analysed through the French medico‐administrative database. Features associated with the presence of ILD were studied. Cox hazard model was used to measure the impact of ILD on survival from the first stay to 2020. The incidence of ILD in SLE was estimated by analysing the onset of ILD from 2013 to 2020 in SLE patients who had no evidence of ILD in 2013.

Results

Between 2011 and 2012, 10,460 SLE patients had at least one hospital stay and could be traced until 2020. Among them, 134 (1.2%) had an ILD diagnosed at baseline. The frequency of ILD in SLE was higher in patients who had an associated autoimmune disease such as Sjögren's syndrome or systemic sclerosis (29.9% vs. 5.9%, p < 0.0001). ILD was associated with an increased risk of death in SLE in the multivariable analysis (hazard ratio [95% CI] 1.992 [1.420–2.794]; p < 0.0001). Among the 31,029 SLE patients with no evidence of ILD at baseline, ILD occurred in 795 (2.6%) between 2013 and 2020. The incidence rate of ILD in SLE was 10.26 for 1000 patient‐years [95% CI: 10.24–10.28].

Conclusion

In SLE, ILD is exceedingly rare, often associated with another systemic autoimmune disorder and appears as a major risk factor for death.

Outcomes in patients with acute myocardial infarction and history of illicit drug use: a French nationwide analysis

AIMS

Several reports suggest that illicit drug use may be a major cause of acute myocardial infarction (AMI) independently of smoking habits and associated with a poorer prognosis. The aim of our study was to evaluate the impact of illicit drug use on (i) the risk of AMI and (ii) its prognosis.

METHODS AND RESULTS

This French longitudinal cohort study was based on the administrative hospital-discharge database from the entire population. First, we collected data for all patients admitted in hospital in 2013 with at least 5 years of follow-up to identify potential predictors of AMI. In a second phase, we collected data for all patients admitted with AMI from January 2010 to December 2018. We identified patients with a history of illicit drug use (cannabis, cocaine, or opioid). These patients were matched with patients without illicit drug use to assess their prognosis. In 2013, 3 381 472 patients were hospitalized with a mean follow-up of 4.7 ± 1.8 years. In multivariable analysis, among all drugs under evaluation, only cannabis use was significantly associated with a higher risk of AMI [HR 1.32 (95% CI 1.09-1.59), P = 0.004]. Between January 2010 and December 2018, we then identified 738 899 AMI patients. Among these patients, 3827 (0.5%) had a known history of illicit drug use. These patients were younger, most often male and had less comorbidities. After 1:1 propensity score matching, during a mean follow-up of 1.9 ± 2.3 years, there was no significant difference between patients without illicit drug use and patients with illicit drug use regarding all-cause death, cardiovascular death, stroke, or heart failure.

CONCLUSION

In a large and systematic nationwide analysis, cannabis use was an independent risk factor for the incidence of AMI. However, the prognosis of illicit drug users presenting with AMI was similar to patients without illicit drug use.

Validation of Cancer Diagnosis Based on the National Health Insurance Service Database versus the National Cancer Registry Database in Korea

Purpose

This study aimed to assess the feasibility of operational definitions of cancer patients in conducting cancer-related studies using the claims data from the National Health Insurance Service (NHIS).

Materials and Methods

Cancer incidence data were obtained from The Korean Central Cancer Registry (KCCR), NHIS primary diagnosis, and from the rare and intractable disease (RID) registration program.

Results

The operational definition with higher sensitivity for cancer patient verification was different by cancer type. Using primary diagnosis, the lowest sensitivity was found in colorectal cancer (91.5%; 95% confidence interval [CI], 91.7 to 92.0) and the highest sensitivity was found in breast cancer (97.9%; 95% CI, 97.8 to 98.0). With RID, sensitivity was the lowest in liver cancer (91.9%; 95% CI, 91.7 to 92.0) and highest in breast cancer (98.1%; 95% CI, 98.0 to 98.2). In terms of the difference in the date of diagnosis in the cancer registration data, > 80% of the patients showed a < 31-day difference from the RID definition.

Conclusion

Based on the health claims data, the operational definition of cancer incidence is more accurate when using the RID registration program claims compared to using the primary diagnosis despite the relatively less concordance by cancer type requires additional definitions such as treatment.

How to produce sound predictions of incidence at a district level using either health care or mortality data in the absence of a national registry: the example of cancer in France

BACKGROUND

In many countries, epidemiological surveillance of chronic diseases is monitored by local registries (LR) which do not necessarily cover the whole national territory. This gap has fostered interest in using non-registry databases (e.g., health care or mortality databases) available for the whole territory as proxies for incidence at the local level. However, direct counts from these databases do not provide reliable incidence measures. Accordingly, specific methods are needed to correct proxies and assess their epidemiological usefulness.

METHODS

This study's objective was to implement a three-stage turnkey methodology using national non-registry data to predict incidence in geographical areas without an LR as follows: constructing a calibration model to make predictions including accurate prediction intervals; accuracy assessment of predictions and rationale for the criteria to assess which predictions were epidemiologically useful; mapping after spatial smoothing of the latter predictions. The methodology was applied to a real-world setting, whereby we aimed to predict cancer incidence, by gender, at the district level in France over the 2007-15 period for 24 different cancer sites, using several health care indicators and mortality. In the present paper, the spatial smoothing performed on predicted incidence of epidemiological interest is illustrated for two examples.

RESULTS

Predicted incidence of epidemiological interest was possible for 27/34 solid site-gender combinations and for only 2/8 haematological malignancies-gender combinations. Mapping of smoothed predicted incidence provided a clear picture of the main contrasts in incidence between districts.

CONCLUSIONS

The methodology implemented provides a comprehensive framework to produce valuable predictions of incidence at a district level, using proxy measures and existing LR.

Validation study: evaluation of the metrological quality of French hospital data for perinatal algorithms

OBJECTIVE

The aim of our validation study was to assess the metrological quality of hospital data for perinatal algorithms on a national level.

DESIGN

Validation study.

SETTING

This was a multicentre study of the French medicoadministrative database on perinatal indicators.

PARTICIPANTS

In each hospital, we selected 150 discharge abstracts for delivery (after 22 weeks of gestation), in 2014, and their corresponding medical records. Overall, 22 hospitals were included.

INTERVENTIONS

A single investigator performed blind data collection from medical records in order to compare data from discharge abstracts with data from medical records. Finally, 3246 discharge abstracts were studied.

PRIMARY AND SECONDARY OUTCOME MEASURES

Seventy items, including maternal and delivery characteristics and maternal morbidity, were collected for each delivery stay.

RESULTS

The concordance rate of maternal age at delivery was 94.8% (95% CI 93.8 to 95.4). Combining the two forms of pre-existing diabetes, the algorithm presented a PPV of 65.9% and a sensitivity of 75.7%. The concordance rate of gestational age at delivery was 91.8% (90.9 to 92.7). Regarding gestational diabetes, the PPV was 80.8% (79.4 to 82.2) and the sensitivity was 79.5% (78.1 to 80.9). Regardless of the algorithm explored, the PPV for vaginal delivery was over 99%. For the diagnosis codes corresponding to immediate postpartum haemorrhage, the PPV was 77.7% (76.3 to 79.1) and the sensitivity was 75.5% (74.0 to 77.0). The algorithm for stillbirth presented a PPV of 89.4% (88.3 to 90.5) and a sensitivity of 95.4% (94.7 to 96.1).

CONCLUSIONS

This first national validation study of many perinatal algorithms suggests that the French national hospital database is an appropriate data source for epidemiological studies, except for some indicators which presented low PPV and/or sensitivity.

Risk of acute arterial events associated with treatment of inflammatory bowel diseases: nationwide French cohort study

OBJECTIVE

Patients with IBD are at increased risk of acute arterial events. Antitumour necrosis factor (TNF) agents and thiopurines may, via their anti-inflammatory properties, lower the risk of acute arterial events. The aim of this study was to assess the impact of thiopurines and anti-TNFs on the risk of acute arterial events in patients with IBD.

DESIGN

Patients aged 18 years or older and affiliated to the French national health insurance with a diagnosis of IBD were followed up from 1 April 2010 until 31 December 2014. The risks of acute arterial events (including ischaemic heart disease, cerebrovascular disease and peripheral artery disease) were compared between thiopurines and anti-TNFs exposed and unexposed patients with marginal structural Cox proportional hazard models adjusting for baseline and time-varying demographics, medications, traditional cardiovascular risk factors, comorbidities and IBD disease activity.

RESULTS

Among 177 827 patients with IBD (96 111 (54%) women, mean age at cohort entry 46.2 years (SD 16.3), 90 205 (50.7%) with Crohn's disease (CD)), 4145 incident acute arterial events occurred (incidence rates: 5.4 per 1000 person-years). Compared with unexposed patients, exposure to anti-TNFs (HR 0.79, 95% CI 0.66 to 0.95), but not to thiopurines (HR 0.93, 95% CI 0.82 to 1.05), was associated with a decreased risk of acute arterial events. The magnitude in risk reduction was highest in men with CD exposed to anti-TNFs (HR 0.54, 95% CI 0.40 to 0.72).

CONCLUSION

Exposure to anti-TNFs is associated with a decreased risk of acute arterial events in patients with IBD, particularly in men with CD.

Detection of incident breast and colorectal cancer cases from an administrative healthcare database in Catalonia, Spain

OBJECTIVE

To validate the Catalan minimum basic data set (MBDS) of hospital discharges as an information source for detecting incident breast (BC) and colorectal cancer (CRC), against the Hospital del Mar Cancer Registry (RTHMar) in Barcelona (Spain) as the gold standard.

METHODS

Using ASEDAT software (Analysis, Selection and Extraction of Tumour Data), we identified Catalan public hospital discharge abstracts in patients with a first-time diagnosis of BC and CRC in the years 2005, 2008, and 2011, aggregated by unique patient identifiers and sorted by date. Once merged with the RTHMar database and anonymized, tumour-specific algorithms were validated to extract data on incident cases, tumour stage, surgical treatment, and date of incidence.

RESULTS

MBDS had a respective sensitivity and positive predictive value (PPV) of 78.0% (564/723) and 90.5% (564/623) for BC case detection; and 83.9% (387/461) and 94.9% (387/408) for CRC case detection. The staging algorithms overestimated the proportion of local-stage cases and underestimated the regional-stage cases in both cancers. When loco-regional stage and surgery were combined, sensitivity and PPV reached 98.3% and 99.8%, respectively, for BC and 96.4% and 98.4% for CRC. The differences between dates of incidence between RTHMar and MBDS were greater for BC cases without initial surgery, whereas they were generally smaller and homogeneous for CRC cases.

CONCLUSIONS

The MBDS is a valid and efficient instrument to improve the completeness of a hospital-based cancer registry (HBCR), particularly in BC and CRC, which require hospitalization and are predominantly surgical.

Combining population-based administrative health records and electronic medical records for disease surveillance

BACKGROUND

Administrative health records (AHRs) and electronic medical records (EMRs) are two key sources of population-based data for disease surveillance, but misclassification errors in the data can bias disease estimates. Methods that combine information from error-prone data sources can build on the strengths of AHRs and EMRs. We compared bias and error for four data-combining methods and applied them to estimate hypertension prevalence.

METHODS

Our study included rule-based OR and AND methods that identify disease cases from either or both data sources, respectively, rule-based sensitivity-specificity adjusted (RSSA) method that corrects for inaccuracies using a deterministic rule, and probabilistic-based sensitivity-specificity adjusted (PSSA) method that corrects for error using a statistical model. Computer simulation was used to estimate relative bias (RB) and mean square error (MSE) under varying conditions of population disease prevalence, correlation amongst data sources, and amount of misclassification error. AHRs and EMRs for Manitoba, Canada were used to estimate hypertension prevalence using validated case definitions and multiple disease markers.

RESULTS

The OR method had the lowest RB and MSE when population disease prevalence was 10%, and the RSSA method had the lowest RB and MSE when population prevalence increased to 20%. As the correlation between data sources increased, the OR method resulted in the lowest RB and MSE. Estimates of hypertension prevalence for AHRs and EMRs alone were 30.9% (95% CI: 30.6-31.2) and 24.9% (95% CI: 24.6-25.2), respectively. The estimates were 21.4% (95% CI: 21.1-21.7), for the AND method, 34.4% (95% CI: 34.1-34.8) for the OR method, 32.2% (95% CI: 31.8-32.6) for the RSSA method, and ranged from 34.3% (95% CI: 34.1-34.5) to 35.9% (95% CI, 35.7-36.1) for the PSSA method, depending on the statistical model.

CONCLUSIONS

The OR and AND methods are influenced by correlation amongst the data sources, while the RSSA method is dependent on the accuracy of prior sensitivity and specificity estimates. The PSSA method performed well when population prevalence was high and average correlations amongst disease markers was low. This study will guide researchers to select a data-combining method that best suits their data characteristics.

C2 HEST Score and Prediction of Incident Atrial Fibrillation in Poststroke Patients: A French Nationwide Study

Background

The C₂HEST score (coronary artery disease or chronic obstructive pulmonary disease [1 point each]; hypertension [1 point]; elderly [age ≥75 years, 2 points]; systolic heart failure [2 points]; thyroid disease [hyperthyroidism, 1 point]) was initially proposed for predicting incident atrial fibrillation (AF) in the general population. Its performance in poststroke patients remains to be established, especially because patients at high risk for incident AF should be targeted for more comprehensive screening. This study aimed to evaluate this newly established incident AF prediction risk score in a post–ischemic stroke population.

Methods and Results

Validation was based on a hospital‐based nationwide cohort with 240 459 French post–ischemic stroke patients. Kaplan–Meier curves for incident rate of AF depict differences between varying risk categories. Discrimination of the C₂HEST score was evaluated using the C index, the net reclassification index, integrated discriminatory improvement, and decision curve analysis. During 7.9±11.5 months of follow‐up, 14 095 patients developed incident AF. The incidence of AF increased from 23.5 per 1000 patient‐years in patients with a C₂HEST score of 0 to 196.8 per 1000 patient‐years in patients with a C₂HEST score ≥6. Kaplan–Meier curves showed a clear difference among different risk strata (log‐rank P<0.0001). The C₂HEST score had good discrimination with a C index of 0.734 (95% CI, 0.732–0.736), which was better than the Framingham risk score and the CHA₂DS₂‐VASc score (congestive heart failure, hypertension, age ≥75 [doubled], diabetes mellitus, stroke [doubled], vascular disease, age 65 to 74 years, and female sex) (P<0.0001, respectively). The C₂HEST score was also superior to the Framingham risk score and the CHA₂DS₂‐VASc score as shown by the net reclassification index, integrated discriminatory improvement (P<0.0001, respectively) and decision curve analysis.

Conclusions

The C₂HEST score performed well in discriminating the individual risk of developing incident AF in a white European population hospitalized with previous ischemic stroke. This simple score may potentially be used as a risk stratification tool for decision making in relation to a screening strategy for AF in post–ischemic stroke patients.

Detection of complicated ectopic pregnancies in the hospital discharge database: A validation study

Objective

Complicated ectopic pregnancies with severe bleeding (CEPSB) are life-threatening situations and should be considered maternal near-miss cases. Previous studies have found an association between severe maternal morbidity secondary to CEPSB and substandard care. Almost all women with CEPSB are hospitalized, generating administrative and medical records. The objective of this study was to propose a method to measure the validity of the hospital discharge database (HDD) to detect CEPSB among hospital stays in two gynecological units.

Methods

We included all hospital stays of women who were 18–45 years old and hospitalized for acute pelvic pain or/and metrorrhagia in the two hospitals. The HDD was compared to medical data (gold standard). Two algorithms constructed from the International Classification of Disease (ICD-10) and Common Classification of Medical Procedures (CCAM), were applied to the HDD: a “predefined algorithm” according to coding guidelines and a “pragmatic algorithm” based on coding practices. Sensitivity, specificity and positive likelihood-ratios were calculated. False negatives and positives were analyzed to describe coding practices.

Results

Among 370 hospital stays included, 52 were classified as CEPSB cases. The “predefined algorithm” gave a sensitivity of 23.1% (95% CI: 11.6–34.5) and a specificity of 99.1% (95% CI: 98.0–100.0) to identify CEPSB. The “pragmatic algorithm” gave a sensitivity of 63.5% (95% CI: 50.4–76.5) and a specificity of 94.7% (95% CI: 92.2–97.5) to identify CEPSB. Coding errors (77.6%) were due to misuse of diagnosis codes and because complications were not coded.

Conclusion

HDD is not reliable enough to detect CEPSB due to incorrect coding practices. However, it could be an ideal tool to monitor quality of care if a culture in data quality assessment is developed to improve quality of medical information.

Leveraging Linkage of Cohort Studies With Administrative Claims Data to Identify Individuals With Cancer

BACKGROUND

In an effort to overcome quality and cost constraints inherent in population-based research, diverse data sources are increasingly being combined. In this paper, we describe the performance of a Medicare claims-based incident cancer identification algorithm in comparison with observational cohort data from the Nurses' Health Study (NHS).

METHODS

NHS-Medicare linked participants' claims data were analyzed using 4 versions of a cancer identification algorithm across 3 cancer sites (breast, colorectal, and lung). The algorithms evaluated included an update of the original Setoguchi algorithm, and 3 other versions that differed in the data used for prevalent cancer exclusions.

RESULTS

The algorithm that yielded the highest positive predictive value (PPV) (0.52-0.82) and κ statistic (0.62-0.87) in identifying incident cancer cases utilized both Medicare claims and observational cohort data (NHS) to remove prevalent cases. The algorithm that only used NHS data to inform the removal of prevalent cancer cases performed nearly equivalently in statistical performance (PPV, 0.50-0.79; κ, 0.61-0.85), whereas the version that used only claims to inform the removal of prevalent cancer cases performed substantially worse (PPV, 0.42-0.60; κ, 0.54-0.70), in comparison with the dual data source-informed algorithm.

CONCLUSIONS

Our findings suggest claims-based algorithms identify incident cancer with variable reliability when measured against an observational cohort study reference standard. Self-reported baseline information available in cohort studies is more effective in removing prevalent cancer cases than are claims data algorithms. Use of claims-based algorithms should be tailored to the research question at hand and the nature of available observational cohort data.

Septic shock among patients with systemic lupus erythematosus: Short and long-term outcome. Analysis of a French nationwide database

OBJECTIVES

We aimed to assess the characteristics, outcomes and costs of septic shock complicating Systemic Lupus Erythematosus (SLE).

METHODS

Characteristics of SLE patients experiencing a septic shock in France from 2010 to 2015 were analyzed through the French medico-administrative database. Factors associated with the 1-year post-admission mortality were analyzed, the crude 1-year survival of SLE patients experiencing septic shock was compared to those admitted for another reason, and we compared the 1-year outcome associated with SLE septic shock survival to a matched SLE ICU control population.

RESULTS

Among 28,522 SLE patients, 1068 experienced septic shock. The 1-year mortality rate was 43.4%. Independently of the severity, an associated Sjögren syndrome was the only specific SLE phenotype associated with mortality (HR 1.392[1.021-1.899]). Within one year, post-septic shock survivors (n = 738) were re-admitted 6.42[17.3] times with total cost of € 14,431[20,444]. Unmatched analysis showed that the outcome of patients admitted in ICU for septic shock was poorer than that of patients admitted in ICU or hospital for another disease. However, 1-year healthcare use of septic shock survivors was not different from the other ICU survivors when matched on severity.

CONCLUSIONS

Septic shock is a frequent and severe complication among SLE patients even if it is not associated with more healthcare use than another episode of same severity.

Accuracy of administrative databases in detecting primary breast cancer diagnoses: a systematic review

OBJECTIVE

To define the accuracy of administrative datasets to identify primary diagnoses of breast cancer based on the International Classification of Diseases (ICD) 9th or 10th revision codes.

DESIGN

Systematic review.

DATA SOURCES

MEDLINE, EMBASE, Web of Science and the Cochrane Library (April 2017).

ELIGIBILITY CRITERIA

The inclusion criteria were: (a) the presence of a reference standard; (b) the presence of at least one accuracy test measure (eg, sensitivity) and (c) the use of an administrative database.

DATA EXTRACTION

Eligible studies were selected and data extracted independently by two reviewers; quality was assessed using the Standards for Reporting of Diagnostic accuracy criteria.

DATA ANALYSIS

Extracted data were synthesised using a narrative approach.

RESULTS

From 2929 records screened 21 studies were included (data collection period between 1977 and 2011). Eighteen studies evaluated ICD-9 codes (11 of which assessed both invasive breast cancer (code 174.x) and carcinoma in situ (ICD-9 233.0)); three studies evaluated invasive breast cancer-related ICD-10 codes. All studies except one considered incident cases.The initial algorithm results were: sensitivity ≥80% in 11 of 17 studies (range 57%-99%); positive predictive value was ≥83% in 14 of 19 studies (range 15%-98%) and specificity ≥98% in 8 studies. The combination of the breast cancer diagnosis with surgical procedures, chemoradiation or radiation therapy, outpatient data or physician claim may enhance the accuracy of the algorithms in some but not all circumstances. Accuracy for breast cancer based on outpatient or physician's data only or breast cancer diagnosis in secondary position diagnosis resulted low.

CONCLUSION

Based on the retrieved evidence, administrative databases can be employed to identify primary breast cancer. The best algorithm suggested is ICD-9 or ICD-10 codes located in primary position.

TRIAL REGISTRATION NUMBER

CRD42015026881.

Increased risk of acute arterial events in young patients and severely active IBD: a nationwide French cohort study

OBJECTIVE

Magnitude and independent drivers of the risk of acute arterial events in IBD are still unclear. We addressed this question in patients with IBD compared with the general population at a nationwide level.

DESIGN

Using the French National Hospital Discharge Database from 2008 to 2013, all patients aged 15 years or older and diagnosed with IBD were identified and followed up until 31 December 2013. The rates of incident acute arterial events were calculated and the impact of time with active disease (period around hospitalisation for IBD flare or IBD-related surgery) on the risk was assessed by Cox regression adjusted for traditional cardiovascular risk factors.

RESULTS

Among 210 162 individuals with IBD (Crohn's disease (CD), n=97 708; UC, n=112 454), 5554 incident acute arterial events were identified. Both patients with CD and UC had a statistically significant overall increased risk of acute arterial events (standardised incidence ratio (SIR) 1.35; 95% CI 1.30 to 1.41 and SIR 1.10; 95 CI 1.06 to 1.13, respectively). The highest risk was observed in patients under the age of 55 years, both in CD and UC. The 3-month periods before and after IBD-related hospitalisation were associated with an increased risk of acute arterial events in both CD and UC (HR 1.74; 95 CI 1.44 to 2.09 and 1.87; 95% CI 1.58 to 2.22, respectively).

CONCLUSION

Patients with IBD are at increased risk of acute arterial events, with the highest risk in young patients. Disease activity may also have an independent impact on the risk.

Prediction of Incident Atrial Fibrillation According to Gender in Patients With Ischemic Stroke From a Nationwide Cohort

The CHA₂DS₂-VASc score may identify patients at higher risk of atrial fibrillation (AF) following ischemic stroke (IS) in patients without known AF. We compared gender-related differences in items from CHA₂DS₂-VASc score and their relation with AF occurrence after IS. This French cohort study was based on the database covering hospital care from 2009 to 2012 for the entire population. Of 336,291 patients with IS, 240,459 (71.5%) had no AF at baseline. Women were older, more frequently had hypertension, heart failure, and had a higher CHA₂DS₂-VASc score than men (4.63 vs 4.39, p<2DS₂-VASc score items were independent predictors of incident AF, except diabetes and vascular disease). Results were mostly similar in men and women when one analyzed separately these predictors. Predictive value of the CHA₂DS₂-VASc score for identifying patients at higher risk of incident AF was somewhat higher in men (C statistic 0.720, 95% confidence interval 0.717 to 0.722) than in women (0.702, 95% confidence interval 0.699 to 0.704). Coronary artery disease, valvular disease, and history of pacemaker or defibrillator implantation were also independent predictors of incident AF. In conclusion, there were significant differences in co-morbidities, possible mechanisms, incidence, and predictors of AF between men and women after IS. However, a strategy using CHA₂DS₂-VASc score for identifying a higher risk of incident AF following IS was useful in both genders.

Hospital Discharge Database as a Tool to Monitor Incidence, Survival and Burden of Cancer in Adolescents and Young Adults

Aims and Background

Cancer in young patients (15–39 years) is unique for the distribution of types, therapeutic options and clinical evolution. Administrative databases represent well-documented tools in epidemiology, and in oncology they are very important in those realities without cancer registries. Our study aimed to analyze the occurrence, outcomes and burden of cancer in young patients through the analysis of hospital discharge records.

Methods

Hospital discharge databases and civil registries were analyzed through record linkage technique. Annual incidence rate (AIR), standardized incidence rate (SR), overall survival, hospitalization rate, and mean number of hospitalizations were evaluated.

Results

Among 2,330,459 young adults, 1846 new cancer patients had been hospitalized in the analyzed period. The SR was 69.3/100,000/year: 1051, 56.9%, were females (AIR 91.0 and SR 76.0) and 795, 43.1%, were males (AIR 67.6 and SR 62.5). Hematological disease was more frequent in males than females (25.5% vs 14.7%, P <0.0001), whereas solid tumors were more frequent among females (85.3% vs 74.5, P <0.0001). The distribution by diagnostic group showed that among females breast cancer was the most frequent (n = 272, SR 17.2), whereas among males genitourinary tract cancer (n = 245, SR 19.2), especially testicular cancer (n = 187, SR 15.1), was the most frequent. Metastatic disease at diagnosis was already present in 198 patients with a solid cancer (13.3%), whereas 213 (11.5%) developed metastasis in the following years. At 12 months from the diagnosis, 87 of 1488 patients with solid cancers and 35 of 358 patients with hematologic disease failed: overall survival was 94% and 90%, respectively. Patients with a new diagnosis of cancer had produced 6663 hospitalizations, 4640 (69.6%) of which were due to solid tumors, 3992 (59.9%) produced by patients over 29 years old, and 3606 (54.1%) by females. The percentage of day hospital admissions increased proportionally with patient age: 25.7% of all hospitalizations among older adolescents (15–20 years) and 32.9% among young adults of 34–39 years.

Conclusions

Administrative data have clear advantages in terms of availability and large numbers. Comparison of our results with the literature showed that a health care delivery database can provide useful information for clinical-epidemiologic evaluations in oncology as well as for the analysis of health services utilization.

Using administrative data to identify and stage breast cancer cases: Implications for assessing quality of care

Aims and background

The study evaluated the use of Italian hospital discharge data (SDO, scheda di dimissione ospedaliera) for identifying women with incident breast cancer, determining stage at diagnosis and assessing quality of care.

Study design

Women aged 20+ years residing in the Regione Emilia-Romagna, Italy, between 2002 and 2005 were studied. Case identification using algorithms based on ICD-9-CM codes on hospital discharge data were compared with AIRTUM-accredited cancer registry data. Sensitivity, specificity and positive predictive value were computed overall, by age and cancer stage. Compliance with guidelines for radiation therapy using registry and hospital data were compared.

Results

A total of 11,615 women was identified by AIRTUM-accredited cancer registries as incident cases, whereas 10,876 women were identified by the SDO algorithm. Sensitivity was 84.8%, specificity was 99.9%, and the positive predictive value was 90.6%. Of the 1,022 who were false positives, 363 (35.5%) were women identified in registry data as having an incident case prior to 2002 and therefore were not included in the analysis. There were 1,761 false negatives; nearly 50% were over 70 years of age or did not undergo a surgical procedure and therefore were not included in our SDO-based case finding. Sensitivity declined as the patient population became older. However, we observed relatively good positive predictive value for all age groups. Algorithms using the SDO data did not clearly identify specific cancer stages. However, the algorithm may have utility where stages are grouped together for use in quality measures.

Conclusions

Cases were identified with good sensitivity, specificity and positive predictive value with SDO data, with better rates than similar previously published algorithms based on Italian data. These hospital claims-based algorithms facilitate quality of care analyses for large populations when registry data are not available by identifying individual women and their subsequent use of health care services.

Evaluation of algorithms to identify incident cancer cases by using French health administrative databases

PURPOSE

Administrative databases are increasingly being used in cancer observational studies. Identifying incident cancer in these databases is crucial. This study aimed to develop algorithms to estimate cancer incidence by using health administrative databases and to examine the accuracy of the algorithms in terms of national cancer incidence rates estimated from registries.

METHODS

We identified a cohort of 463 033 participants on 1 January 2012 in the Echantillon Généraliste des Bénéficiaires (EGB; a representative sample of the French healthcare insurance system). The EGB contains data on long-term chronic disease (LTD) status, reimbursed outpatient treatments and procedures, and hospitalizations (including discharge diagnoses, and costly medical procedures and drugs). After excluding cases of prevalent cancer, we applied 15 algorithms to estimate the cancer incidence rates separately for men and women in 2012 and compared them to the national cancer incidence rates estimated from French registries by indirect age and sex standardization.

RESULTS

The most accurate algorithm for men combined information from LTD status, outpatient anticancer drugs, radiotherapy sessions and primary or related discharge diagnosis of cancer, although it underestimated the cancer incidence (standardized incidence ratio (SIR) 0.85 [0.80-0.90]). For women, the best algorithm used the same definition of the algorithm for men but restricted hospital discharge to only primary or related diagnosis with an additional inpatient procedure or drug reimbursement related to cancer and gave comparable estimates to those from registries (SIR 1.00 [0.94-1.06]).

CONCLUSION

The algorithms proposed could be used for cancer incidence monitoring and for future etiological cancer studies involving French healthcare databases. Copyright © 2017 John Wiley & Sons, Ltd.

Assessment of the French National Health Insurance Information System as a tool for epidemiological surveillance of malaria

OBJECTIVE

Epidemiological surveillance of malaria in France is based on a hospital laboratory sentinel surveillance network. There is no comprehensive population surveillance. The objective of this study was to assess the ability of the French National Health Insurance Information System to support nationwide malaria surveillance in continental France.

MATERIALS AND METHODS

A case identification algorithm was built in a 2-step process. First, inclusion rules giving priority to sensitivity were defined. Then, based on data description, exclusion rules to increase specificity were applied. To validate our results, we compared them to data from the French National Reference Center for Malaria on case counts, distribution within subgroups, and disease onset date trends.

RESULTS

We built a reusable automatized tool. From July 1, 2013, to June 30, 2014, we identified 4077 incident malaria cases that occurred in continental France. Our algorithm provided data for hospitalized patients, patients treated by private physicians, and outpatients for the entire population. Our results were similar to those of the National Reference Center for Malaria for each of the outcome criteria.

DISCUSSION

We provided a reliable algorithm for implementing epidemiological surveillance of malaria based on the French National Health Insurance Information System. Our method allowed us to work on the entire population living in continental France, including subpopulations poorly covered by existing surveillance methods.

CONCLUSION

Traditional epidemiological surveillance and the approach presented in this paper are complementary, but a formal validation framework for case identification algorithms is necessary.

[The human and economic burden of cancer in France in 2014, based on the Sniiram national database]

INTRODUCTION

The national health insurance information system (Sniiram) can be used to estimate the national medical and economic burden of cancer. This study reports the annual rates, characteristics and expenditure of people reimbursed for cancer.

METHODS

Among 57 million general health scheme beneficiaries (86% of the French population), people managed for cancer were identified using algorithms based on hospital diagnoses and full refund for long-term cancer. The reimbursed costs (euros) related to the cancer, paid off by the health insurance, were estimated.

RESULTS

In 2014, 2.491 million people (4.4%) covered by the general health scheme had a cancer managed (men 1.1 million, 5.1%; women 1.3 million, 4.9%). The annual (2012-2014) average growth rate of patients was 0.8%. The spending related to the cancer was 13.5 billion: 5 billion for primary health care (drugs 2.3 billion), 7.5 billion for the hospital (drugs 1.3 billions) and 900 million for sick leave and invalidity pensions. Spending annual average growth rate (2012-2014) was 4% (drugs 2%). The rates of patients and the relative spending were 1.8% and 2.5 billion for the breast cancer (women), 1.5% and 1.0 billion for prostate cancer, 0.9% and 1.5 billion for the colon cancer, and 0.19% and 1.3 billion for lung cancer.

DISCUSSION

Cancers establish one of the first groups of chronic diseases pathologies in terms of patients and spending. If the numbers of patients remain stables, the spending increases, mainly for medicines.

Geographical variations in incidence, management and survival of hepatocellular carcinoma in a Western country

BACKGROUND & AIMS

Information on the incidence, management, and prognosis of hepatocellular carcinoma (HCC) is derived from population samples, regional data, or registries. Comprehensive national evaluations within a given country are lacking. This study aimed to investigate regional variations in HCC care within France.

METHODS

This observational study analysed data from French administrative databases for more than 30,000 patients with HCC diagnosed between 2009 and 2012, and followed-up until 2013. The incidence of HCC, access to surgery, and survival, at both the national level and two geographical levels (the 21 French regions and 95 French departments into which France is divided administratively), were determined. The influence on outcome of the structure of the hospital where HCC was first managed was assessed.

RESULTS

At the national level, the median survival was 9.4months and only 22.8% of patients had curative treatment. There were marked variations between regions and departments in incidence, access to curative treatment (range 1.3-28.8% and 8.1-32.3% respectively), and in median survival (range 5.7-12.1 and 4.3-16.5months respectively). The administrative type and annual HCC-caseload of the hospital where patients were first admitted also had an independent influence on treatment and survival.

CONCLUSION

Despite full insurance coverage for all citizens, national measures to reduce inequities in the management of cancer patients, standardised recommendations for HCC surveillance and management, the percentage of patients undergoing curative treatment and their survival may vary four-fold depending on their postcode. The hospital in which patients are first managed has a clear influence on accessibility to both good care and survival.

LAY SUMMARY

Population-based studies have highlighted large and sometimes unexpected differences between countries in the survival of patients with malignancy. As these differences are considered to indicate the overall effectiveness of health systems, in addition to the incidence of the cancer or quality of registration, variations within a given country should be minimal. However, similar to between countries differences, this study shows differences within the same country in the incidence, curative treatment rate, and survival of patients with HCC. Evidence that access to care and survival varies within a country can strengthen the impetus for government and clinicians to address these disparities.

Geographic variation in potentially avoidable hospitalizations in France

Potentially avoidable hospitalizations are studied as an indirect measure of access to primary care. Understanding the determinants of these hospitalizations can help improve the quality, efficiency, and equity of health care delivery. Few studies have tackled the issue of potentially avoidable hospitalizations in France, and none has done so at the national level. We assessed disparities in potentially avoidable hospitalizations in France in 2012 and analyzed their determinants. The standardized rate of potentially avoidable hospitalizations ranged from 0.1 to 44.4 cases per 1,000 inhabitants, at the ZIP code level. Increased potentially avoidable hospitalizations were associated with higher mortality, lower density of acute care beds and ambulatory care nurses, lower median income, and lower education levels. This study unveils considerable variation in the rate of potentially avoidable hospitalizations in spite of France's mandatory, publicly funded health insurance system. In addition to epidemiological and sociodemographic factors, this study suggests that primary care organization plays a role in geographic disparities in potentially avoidable hospitalizations that might be addressed by increasing the number of nurses and enhancing team work in primary care. Policy makers should consider measuring potentially avoidable hospitalizations in France as an indicator of primary care organization.

Predictive ability of HAS-BLED, HEMORR2HAGES, and ATRIA bleeding risk scores in patients with atrial fibrillation. A French nationwide cross-sectional study

OBJECTIVE

The HAS-BLED, ATRIA, and HEMORR2HAGES risk scores were created to evaluate individual bleeding risk in atrial fibrillation (AF). We sought to estimate and compare the predictive ability of these scores for major hemorrhage in AF, including elderly (≥80years) and non-elderly (<80years) patients.

METHODS

This cross-sectional study is based on the French National Hospital Database (PMSI), which covers the entire French population. Data from all patients with an AF diagnosis in 2012 were extracted. Demographic and comorbidity data were used to calculate the three bleeding risk scores for each patient. Patients hospitalized with a principal diagnosis of major bleeding were identified.

RESULTS

Of the 533,044 AF patients identified, 53.2% were ≥80years; 7013 patients (1.3%) were hospitalized for a bleeding event (1785 for intracranial hemorrhage). Bleeding occurred more frequently in patients with higher HAS-BLED, HEMORR2HAGES, and ATRIA scores. In patients ≥80years, the c-statistics did not differ (p=0.27) between HAS-BLED (0.54; 95% confidence interval [CI]: 0.53-0.54), HEMORR2HAGES (0.53; 95% CI: 0.53-0.54), and ATRIA (0.53; 95% CI: 0.52-0.54). In patients <80years, HAS-BLED (0.59; 95% CI: 0.58-0.60) had a slightly higher c-statistic than HEMORR2HAGES (0.56; 95% CI: 0.55-0.57) and ATRIA (0.55, 95% CI: 0.55-0.56) (p<0.0001).

CONCLUSIONS

Given its simplicity and similar performance, HAS-BLED may be an attractive alternative to HEMORR2HAGES for estimation of bleeding risk in AF patients <80years. However, accurate determination of bleeding risk among the elderly is difficult with existing risk-prediction scores, indicating a clear need for improvement in their clinical utility.

Cause of death in patients with atrial fibrillation admitted to French hospitals in 2012: a nationwide database study

Objective

Most patients with atrial fibrillation (AF) have risk factors and coexisting conditions that increase their mortality risk. We performed a cause-of-death analysis to identify predictors of mortality in hospitalised patients with AF in France.

Methods and results

In this retrospective, population-based cross-sectional study, the Programme de médicalisation des systèmes d'information was used to identify 533 044 adults with a diagnosis of AF or atrial flutter hospitalised for any reason in France from January through December 2012. Stepwise multivariable analyses were performed to identify determinants of mortality. The mean age was 78.0±11.4 years, 47.1% were women, and the mean CHA₂DS₂-VASc score was 4.0±1.8. During hospitalisation, 9.4% (n=50 165) of the patients died, 34% due to a cardiovascular event, most often heart failure (16.6%), stroke/transient ischaemic attack/systemic embolism (9.8%) or vascular or ischaemic disease (4.0%). The strongest predictors of overall death were age ≥75 years (OR 2.57, 95% CI 2.47 to 2.68), renal failure (OR 1.85, 95% CI 1.81 to 1.89), cancer (OR 1.81, 95% CI 1.78 to 1.85) and lung disease (OR 1.58, 95% CI 1.55 to 1.62).

Conclusions

Cardiovascular events were the most common cause of death, occurring in one-third of patients, in this comprehensive study of hospitalised patients with AF. Despite the high risk of stroke in this population, only 10% died from stroke/transient ischaemic attack/systemic embolism. The strongest predictors of overall death were non-cardiovascular. Physicians should be encouraged to focus on preventable serious and disabling cardiovascular events (such as stroke) as well as on potentially fatal non-cardiovascular comorbidities.

Acute Cholecystitis-Optimal Timing for Early Cholecystectomy: a French Nationwide Study

BACKGROUND

The recommended treatment for acute calculous cholecystitis combines antibiotics and cholecystectomy. To reduce morbidity and mortality, guidelines recommend early cholecystectomy. However, the optimal timing for surgery on first admission remains controversial. This study aims to determine the best timing for cholecystectomy in patients presenting with acute calculous cholecystitis.

STUDY DESIGN

The French national health-care database was analyzed to identify all patients undergoing cholecystectomy for acute cholecystitis during the same hospital stay between January 2010 and December 2013. Data regarding patients, procedures, and hospitals characteristics were collected. The relationship between surgery's timing and clinical outcome was evaluated by multiple logistic regressions.

RESULTS

Overall, 42,452 patients from 507 hospitals were included in the study. Postoperative complications requiring invasive treatment occurred in 961 patients (2.3 %), and the mortality rate was 1.1 %. Adverse postoperative outcomes-intensive care admission, reoperation, and postoperative sepsis-were significantly lower when surgery was performed between days 1 and 3 (3-3.3, 0.5-0.6, and 3.8-4.1 %, respectively) when compared to surgery performed on the day of admission (5.6, 1.2, and 5.2 %, p < 0.001) or from day 5 onward (4.5, 1, and 6.5 %, respectively; p < 0.001). Mortality was also significantly lower in patients undergoing cholecystectomy between days 1 and 3 after admission (0.8-1 %) when compared to patients operated on the day of admission or after day 3 (1.4 % on day 0, 1.2 % on day 4, and 1.9 % from day 5: all p < 0.001).

CONCLUSION

For patients with acute calculous cholecystitis, all efforts should be made to perform cholecystectomy within 3 days after hospital admission in order to decrease morbidity and mortality.

Increased Mortality for Elective Surgery during Summer Vacation: A Longitudinal Analysis of Nationwide Data

Surgical safety during vacation periods may be influenced by the interplay of several factors, including workers' leave, hospital activity, climate, and the variety of patient cases. This study aimed to highlight an annually recurring peak of surgical mortality during summer in France and explore its main predictors. We selected all elective of open surgical procedures performed in French hospitals between 2007 and 2012. Surgical mortality variation was analyzed over time in relation to workers leaving on vacation, the volume of procedures performed by hospitals, and temperature changes. We ran a multilevel logistic regression for exploring the determinants of surgical mortality, taking into account the clustering of patients within hospitals and adjusting for patient and hospital characteristics. A total of 609 French hospitals had 8,926,120 discharges related to open elective surgery. During 6 years, we found a recurring mortality peak of 1.15% (95% CI 1.09–1.20) in August compared with 0.81% (0.79–0.82, p<.001) in other months. The incidence of worker vacation was 43.0% (38.9–47.2) in August compared with 7.3% (4.6–10.1, p<.001) in other months. Hospital activity decreased substantially in August (78,126 inpatient stays, 75,298–80,954) in relation to other months (128,142, 125,697–130,586, p<.001). After adjusting for all covariates, we found an "August effect" reflecting a higher risk to patients undergoing operations at this time (OR 1.16, 95% CI 1.12–1.19, p<.001). The main study limitation was the absence of data linkage between surgical staffing and mortality at the hospital level. The observed, recurring mortality peak in August raises questions about how to maintain hospital activity and optimal staffing through better regulation of human activities.

Patients With Ischemic Stroke and Incident Atrial Fibrillation: A Nationwide Cohort Study

BACKGROUND AND PURPOSE

A substantial part of ischemic strokes is attributed to atrial fibrillation (AF). We hypothesized that patients with ischemic stroke without prior diagnosed AF were at higher risk of having a subsequent diagnosis of AF, and this was associated with multiple risk factors.

METHODS

This French longitudinal cohort study was based on the national database covering hospital care from 2008 to 2012 for the entire population.

RESULTS

Of 65 807 patients with ischemic stroke in 2009, 48 992 did not have AF at baseline. A total of 4828 of these patients were diagnosed as having AF during a follow-up of 15±15 months (incidence rate 7.9 per 100 person-years). By comparison, the yearly rate of new-onset AF for the 826 416 patients with a cardiac hospitalization was 5.9%. CHADS2 (congestive heart failure, hypertension, age ≥75 years, diabetes mellitus, stroke/transient ischemic attack) and CHA2DS2-VASc (congestive heart failure, hypertension, age ≥75 years, diabetes mellitus, stroke/transient ischemic attack [doubled], vascular disease, age 65-75 years, and sex category [female]) scores were both associated with the risk of new-onset AF during follow-up (CHADS2: hazard ratio [HR] 1.70, 95% confidence interval [CI] 1.66-1.75; CHA2DS2-VASc: HR 1.45, 95% CI 1.42-1.48). The c statistics were 0.700 (95% CI 0.696-0.706) for CHADS2 and 0.706 (95% CI 0.702-0.710) with CHA2DS2-VASc (P=0.003 for comparison of the 2 scores). Independent predictors of subsequent diagnosis of AF were age 65 to 74 years (HR 2.29, 95% CI 2.06-2.54), age ≥75 years (HR 3.31, 95% CI 3.02-3.64), hypertension (HR 1.22, 95% CI 1.13-1.32), heart failure (HR 2.56, 95% CI 2.41-2.72), and vascular disease (HR 1.10, 95% CI 1.04-1.17).

CONCLUSIONS

Ischemic stroke was associated with a substantially increased risk of incident AF, particularly among individuals with higher CHADS2 or CHA2DS2-VASc scores. These risk scores seem to be simple tools for identifying patients at higher risk of incident AF after ischemic stroke.

Comparison of two methods to report potentially avoidable hospitalizations in France in 2012: a cross-sectional study

Background

Potentially avoidable hospitalizations represent an indirect measure of access to effective primary care. However many approaches have been proposed to measure them and results may differ considerably. This work aimed at examining the agreement between the Weissman and Ansari approaches in order to measure potentially avoidable hospitalizations in France.

Methods

Based on the 2012 French national hospital discharge database (Programme de Médicalisation des Systèmes d’Information), potentially avoidable hospitalizations were measured using two approaches proposed by Weissman et al. and by Ansari et al. Age- and sex-standardised rates were calculated in each department. The two approaches were compared for diagnosis groups, type of stay, severity, age, sex, and length of stay.

Results

The number and age-standardised rate of potentially avoidable hospitalizations estimated by the Weissman et al. and Ansari et al. approaches were 742,474 (13.3 cases per 1,000 inhabitants) and 510,206 (9.0 cases per 1,000 inhabitants), respectively. There are significant differences by conditions groups, age, length of stay, severity level, and proportion of medical stays between the Weissman and Ansari methods.

Conclusions

Regarding potentially avoidable hospitalizations in France in 2012, the agreement between the Weissman and Ansari approaches is poor. The method used to measure potentially avoidable hospitalizations is critical, and might influence the assessment of accessibility and performance of primary care.

Clinical and histopathological factors associated with Ki-67 expression in breast cancer patients

The aim of the present study was to identify the optimal Ki-67 cut-off value in breast cancer (BC) patients, and investigate the association of Ki-67 expression levels with other prognostic factors. Firstly, a retrospective search was performed to identify patients with stage I–III BC (n=462). A range of Ki-67 index values were then assigned to five groups (<10, 10–14, 15–19, 20–24 and ≥25%). The correlation between the Ki-67 index and other prognostic factors [age, tumor type, histological and nuclear grade, tumor size, multifocality, an in situ component, lymphovascular invasion (LVI), estrogen and progesterone receptor (ER/PR) expression, human epidermal growth factor receptor (HER-2) status, axillary involvement and tumor stage] were investigated in each group. The median Ki-67 value was revealed to be 20% (range, 1–95%). A young age (≤40 years old), tumor type, size and grade, LVI, ER/PR negativity and HER-2 positivity were revealed to be associated with the Ki-67 level. Furthermore, Ki-67 was demonstrated to be negatively correlated with ER/PR expression (P<0.001), but positively correlated with tumor size (P<0.001). The multivariate analysis revealed that a Ki-67 value of ≥15% was associated with the largest number of poor prognostic factors (P=0.036). In addition, a Ki-67 value of ≥15% was identified to be statistically significant in association with certain luminal subtypes. The rate of disease-free survival was higher in patients with luminal A subtype BC (P=0.036). Following the correlation analysis for the Ki-67 index and the other prognostic factors, a Ki-67 value of ≥15% was revealed to be the optimal cut-off level for BC patients.

Self-controlled case series and misclassification bias induced by case selection from administrative hospital databases: application to febrile convulsions in pediatric vaccine pharmacoepidemiology

Vaccine safety studies are increasingly conducted by using administrative health databases and self-controlled case series designs that are based on cases only. Often, several criteria are available to define the cases, which may yield different positive predictive values, as well as different sensitivities, and therefore different numbers of selected cases. The question then arises as to which is the best case definition. This article proposes new methodology to guide this choice based on the bias of the relative incidence and the power of the test. We apply this methodology in a validation study of 4 nested algorithms for identifying febrile convulsions from the administrative databases of 10 French hospitals. We used a sample of 695 children aged 1 month to 3 years who were hospitalized in 2008-2009 with at least 1 diagnosis code of febrile convulsions. The positive predictive values of the algorithms ranged from 81% to 98%, and their sensitivities were estimated to be 47%-99% in data from 1 large hospital. When applying our proposed methods, the algorithm we selected used a restricted diagnosis code and position on the discharge abstract. These criteria, which resulted in the selection of 502 cases with a positive predictive value of 95%, provided the best compromise between high power and low relative bias.

Outcomes After Acute Myocardial Infarction in HIV-Infected Patients

Background—

We aimed to assess in-hospital case fatality and 1-year prognosis in HIV-infected patients with acute myocardial infarction.

Methods and Results—

From the PMSI (Program de Medicalisation des Systèmes d’informatique) database, data from 277 303 consecutive acute myocardial infarction patients hospitalized from January 1, 2005, to December 31, 2009, were analyzed. Surviving patients were followed up for 1 year after discharge. HIV-infected patients were compared with uninfected patients. Among the cohort, HIV-infected patients (n=608) accounted for 0.22%. All-cause hospital and 1-year mortality rates were lower in the HIV-infected group than in uninfected patients (3.1% versus 8.1% [ P <0.001] and 1.4% versus 5.5% [ P <0.001], respectively). From the database, we then analyzed a cohort derived from a matching procedure, with 1 HIV patient matched with 2 patients without HIV, based on age and sex (n=1824). Ischemic cardiomyopathy was more frequent in the HIV group (7.6% versus 4.2%, P =0.003). Hospitalization and 1-year mortality rates were similar in the 2 groups (3.1% versus 2.1% [ P =0.168] and 1.4% versus 1.7% [ P =0.642], respectively). However, at 12 months, hospitalizations for episodes of heart failure were significantly more frequent in HIV-infected than in uninfected patients (3.3% versus 1.4%, respectively; P =0.020). HIV infection, diabetes mellitus, history of ischemic cardiomyopathy, and undergoing percutaneous coronary intervention were associated in univariate analysis with occurrence of heart failure. By multivariable analysis, HIV infection (odds ratio 2.82, 95% confidence interval 1.32–6.01), diabetes mellitus, and undergoing percutaneous coronary intervention remained independent predictors of heart failure.

Conclusions—

The present study demonstrates that after acute myocardial infarction, HIV status influences long-term risk, although the short-term risk in HIV patients is comparable to that in uninfected patients.

Automated selection of relevant information for notification of incident cancer cases within a multisource cancer registry

OBJECTIVE

The aim of this study was to develop and evaluate a selection algorithm of relevant records for the notification of incident cases of cancer on the basis of the individual data available in a multi-source information system.

METHODS

This work was conducted on data for the year 2008 in the general cancer registry of Poitou-Charentes region (France). The selection algorithm hierarchizes information according to its level of relevance for tumoral topography and tumoral morphology independently. The selected data are combined to form composite records. These records are then grouped in respect with the notification rules of the International Agency for Research on Cancer for multiple primary cancers. The evaluation, based on recall, precision and F-measure confronted cases validated manually by the registry's physicians with tumours notified with and without records selection.

RESULTS

The analysis involved 12,346 tumours validated among 11,971 individuals. The data used were hospital discharge data (104,474 records), pathology data (21,851 records), healthcare insurance data (7508 records) and cancer care centre's data (686 records). The selection algorithm permitted performances improvement for notification of tumour topography (F-measure 0.926 with vs. 0.857 without selection) and tumour morphology (F-measure 0.805 with vs. 0.750 without selection).

CONCLUSION

These results show that selection of information according to its origin is efficient in reducing noise generated by imprecise coding. Further research is needed for solving the semantic problems relating to the integration of heterogeneous data and the use of non-structured information.

Incidence and prevalence of juvenile idiopathic arthritis among children in a managed care population, 1996-2009

OBJECTIVE

Few studies based in well-defined North American populations have examined the occurrence of juvenile idiopathic arthritis (JIA), and none has been based in an ethnically diverse population. We used computerized healthcare information from the Kaiser Permanente Northern California membership to validate JIA diagnoses and estimate the incidence and prevalence of the disease in this well-characterized population.

METHODS

We identified children aged ≤ 15 years with ≥ 1 relevant International Classification of Diseases, 9th edition, diagnosis code of 696.0, 714, or 720 in computerized clinical encounter data during 1996-2009. In a random sample, we then reviewed the medical records to confirm the diagnosis and diagnosis date and to identify the best-performing case-finding algorithms. Finally, we used the case-finding algorithms to estimate the incidence rate and point prevalence of JIA.

RESULTS

A diagnosis of JIA was confirmed in 69% of individuals with at least 1 relevant code. Forty-five percent were newly diagnosed during the study period. The age- and sex-standardized incidence rate of JIA per 100,000 person-years was 11.9 (95% CI 10.9-12.9). It was 16.4 (95% CI 14.6-18.1) in girls and 7.7 (95% CI 6.5-8.9) in boys. The peak incidence rate occurred in children aged 11-15 years. The prevalence of JIA per 100,000 persons was 44.7 (95% CI 39.1-50.2) on December 31, 2009.

CONCLUSION

The incidence rate of JIA observed in the Kaiser Permanente population, 1996-2009, was similar to that reported in Rochester, Minnesota, USA, but 2 to 3 times higher than Canadian estimates.

Estimation of national colorectal-cancer incidence using claims databases

Background. The aim of the study was to assess the accuracy of the colorectal-cancer incidence estimated from administrative data. Methods. We selected potential incident colorectal-cancer cases in 2004-2005 French administrative data, using two alternative algorithms. The first was based only on diagnostic and procedure codes, whereas the second considered the past history of the patient. Results of both methods were assessed against two corresponding local cancer registries, acting as “gold standards.” We then constructed a multivariable regression model to estimate the corrected total number of incident colorectal-cancer cases from the whole national administrative database. Results. The first algorithm provided an estimated local incidence very close to that given by the regional registries (646 versus 645 incident cases) and had good sensitivity and positive predictive values (about 75% for both). The second algorithm overestimated the incidence by about 50% and had a poor positive predictive value of about 60%. The estimation of national incidence obtained by the first algorithm differed from that observed in 14 registries by only 2.34%. Conclusion. This study shows the usefulness of administrative databases for countries with no national cancer registry and suggests a method for correcting the estimates provided by these data.