Direct costs of adult traumatic spinal cord injury in Ontario.

STUDY DESIGN: Retrospective economic analysis. OBJECTIVES: To determine the total direct costs of publicly funded health care utilization for the three fiscal years 2003/04 to 2005/06 (1 April 2003 to 31 March 2004 to 1 April 2005 to 31 March 2006), from the time of initial hospitalization to 1 year after initial acute discharge among individuals with traumatic spinal cord injury (SCI). SETTING: Ontario, Canada. METHODS: Health system costs were calculated for 559 individuals with traumatic SCI (C1-T12 AIS A-D) for acute inpatient, emergency department, inpatient rehabilitation (that is, short-stay inpatient rehabilitation), complex continuing care (CCC) (i.e., long-stay inpatient rehabilitation), home care services, and physician visits in the year after index hospitalization. All care costs were calculated from the government payer's perspective, the Ontario Ministry of Health and Long-Term Care. RESULTS: Total direct costs of health care utilization in this traumatic SCI population (including the acute care costs of the index event and inpatient readmission in the following year after the index discharge) were substantial: $102 900 per person in 2003/04, $100 476 in 2004/05 and $123 674 in 2005/06 Canadian Dollars (2005 CDN $). The largest cost driver to the health care system was inpatient rehabilitation care. From 2003/04 to 2005/06, the average per person cost of rehabilitation was approximately three times the average per person costs of inpatient acute care. CONCLUSION: The high costs and long length of stay in inpatient rehabilitation are important system cost drivers, emphasizing the need to evaluate treatment efficacy and subsequent health outcomes in the inpatient rehabilitation setting.

FRAX® probabilities and risk of major osteoporotic fracture in France.

UNLABELLED: The incidence of hip fracture, death and the estimated incidence of major osteoporotic fracture in France were used to determine the lifetime and 10-year probability of fracture and incorporated into a probability model (FRAX®) calibrated to the French population. INTRODUCTION: Fracture probabilities in the French population have not been determined. Our aim was to determine the incidence of hip fracture in France and the estimated 10-year probabilities of hip and major osteoporotic fractures. METHODS: The study population included adults over 50 years living in France in 2004. Incident hip fracture cases were identified from the French PMSI database. Incidence of the other major osteoporotic fractures was imputed from the relationship between hip fracture incidence and other major fracture in Sweden. These data were used to calculate population-based fracture probabilities according to age and BMD using cutoff values for femoral neck T-scores from the NHANES III data in Caucasian women. The probability model (FRAX®) calibrated to the French population was used to compute individual fracture probabilities according to specific clinical risk factors. RESULTS: We identified 15,434 men and 51,469 women with an incident hip fracture. The remaining lifetime probability of hip fracture at 50 years was approximately 10 and 30% respectively. With a femoral neck T-score of -2 SD, one in two women and one in five men would sustain a major osteoporotic fracture in their lifetime. The 10-year probability of other major osteoporotic fractures increased with declining T-score and increasing age. Low body mass index and other clinical risk factors had an independent effect on fracture probability whether or not BMD was included in the FRAX® model. CONCLUSION: This analysis provides detailed estimation on the risk of fracture in the French population and may help to define therapeutic guidelines.

[ICD-10 adaptation of 15 Agency for Healthcare Research and Quality patient safety indicators]. / Adaptation au codage CIM-10 de 15 indicateurs de la sécurité des patients proposés par l'Agence étasunienne pour la recherche et la qualité des soins de santé (AHRQ).

BACKGROUND: In the United States, the Agency for Healthcare Research and Quality (AHRQ) has developed 20 Patient Safety Indicators (PSIs) to measure the occurrence of hospital adverse events from medico-administrative data coded according to the ninth revision of the international classification of disease (ICD-9-CM). The adaptation of these PSIs to the WHO version of ICD-10 was carried out by an international consortium. METHODS: Two independent teams transcoded ICD-9-CM diagnosis codes proposed by the AHRQ into ICD-10-WHO. Using a Delphi process, experts from six countries evaluated each code independently, stating whether it was "included", "excluded" or "uncertain". During a two-day meeting, the experts then discussed the codes that had not obtained a consensus, and the additional codes proposed. RESULTS: Fifteen PSIs were adapted. Among the 2569 proposed diagnosis codes, 1775 were unanimously adopted straightaway. The 794 remaining codes and 2541 additional codes were discussed. Three documents were prepared: (1) a list of ICD-10-WHO codes for the 15 adapted PSIs; (2) recommendations to the AHRQ for the improvement of the nosological frame and the coding of PSI with ICD-9-CM; (3) recommendations to the WHO to improve ICD-10. CONCLUSIONS: This work allows international comparisons of PSIs among the countries using ICD-10. Nevertheless, these PSIs must still be evaluated further before being broadly used.

Falls' and fallers' profiles.

OBJECTIVE: To describe the different falls typology and to investigate whether different falls profiles and faller profiles could be identified among a cohort of community-dwelling women aged 75 years and older. DESIGN: Prospective cohort study. PARTICIPANTS: Women aged 75 years and older were enrolled in five French centers after a random selection from electoral lists and included in the EPIDOS study. MEASUREMENTS: During a 4 year follow-up, women were contacted by telephone every 4 months to investigate the occurrence of falls. To minimize the memory bias, the specific questionnaire on falls was completed only if the fall took place in the week preceding the contact. A multiple correspondence analysis followed by clustering was carried out to identify the typology of falls. RESULTS: 727 women described at least one fall. A full description of 662 falls was obtained during the follow-up period. In the multiple correspondence analysis the main discriminant item was outside versus inside falls. Moreover, four clusters were showed: outside falls linked to lack of attention (28% of women), outside falls related to exogenous/environmental factors (16%), inside falls associated with frailty (44%) and in height falls (11%). We also found that each type of falls was correlated with particular health or functional status (i.e.;dependence, motion difficulty, weakness, use of walking aid, ...). CONCLUSION: From frailty to hyperactivity there are different falls and fallers profiles. Assessing such fall profiles could be helpful to develop new dedicated fall prevention programs in the elderly.

Burden of hip fracture on inpatient care: a before and after population-based study.

SUMMARY: We estimated the excess hospital expenditure attributable to osteoporotic hip fracture (HF) within a population of 6,019 patients. Post-fracture excess of hospital days was 23.1, including 22.7 days in rehabilitation care. HF might result from a patient's pre-fracture poor health status rather than predispose to a worsening of such pre-existing conditions. INTRODUCTION: Hip fracture represents a large burden on hospital services. It is unclear whether the post-fracture expenditure is linked to a worsening of pre-fracture comorbid conditions. We estimated the excess hospital expenditure attributable to osteoporotic HF following the initial hospitalization for acute care (index stay). METHODS: We identified 6,019 patients (> or = 50 years) who experienced HF in 2005 and compared their hospitalizations 1 year before and 1 year after the index stay. Excess expenditure was estimated by subtracting the utilization of hospital days or costs (Euros 2005) before the index stay from those after the index stay. Factors associated with hospitalization during the pre-fracture and post-fracture years were identified using multivariate logistic regressions. RESULTS: Beside the index stay, post-fracture excess of hospital days was 23.1 (95% Confidence Interval (CI) [21.8-24.3]), including 22.7 days (95% CI [21.7-23.7]) in rehabilitation care and 0.3 days (95% CI [0-0.9]) in acute care. Estimated excess cost per patient was 5,986 (95% CI [5,638-6,335]) after the index stay, including 5,673 (95% CI [5,419-5,928]) in rehabilitation care. Male and elderly patients were at higher risk to be hospitalized in acute care during the year preceding and succeeding HF. CONCLUSIONS: Osteoporotic HF represents a pronounced excess expenditure in hospital, which is mostly linked to rehabilitation care. Considering that utilization of inpatient acute care was quite similar before and after the index stay, HF might result from a patient's pre-fracture poor health status, rather than predispose to a worsening of such pre-existing conditions.

Characteristics of adults with incident traumatic spinal cord injury in Ontario, Canada.

STUDY DESIGN: Cohort study. OBJECTIVES: To provide recent estimates of the incidence of traumatic spinal cord injury (SCI) in adults living in Ontario. SETTING: Ontario, Canada. METHODS: The study included all men and women aged 18 years and older living in Ontario. The two primary data sources used for this study were the census data provided by Statistics Canada and the hospital Discharge Abstract Database (DAD) provided by the Canadian Institute for Health Information. Incidence was estimated for the fiscal years 2003/04-2006/07, and examined by age, gender, mechanism and seasonality of injury, the level of injury, the presence of comorbidity and in-hospital mortality. RESULTS: The incident cases had a mean age of 51.3 years (s.d. 20.1). The majority of the cases was male (74.1%) and had a cervical SCI caused by falls (49.5%). The age-adjusted incidence rate was stable over the 4-year study period, from 24.2 per million (95% CI: 21.2-27.6) in 2003 to 23.1 per million (95% CI: 20.2-26.3) in 2006. CONCLUSION: Despite worldwide trends that have indicated motor vehicle collisions (MVCs) as the leading cause of injury, falls emerged as the leading cause of traumatic SCI in this study. This finding, and the fact that the number of fall-induced injuries increased steadily with age, may indicate that there is growing concern for the consequences of falls in the elderly. Further work is needed to understand this trend in age and gender and the causes of falls to develop effective fall prevention strategies.

Health care utilization in non-traumatic and traumatic spinal cord injury: a population-based study.

STUDY DESIGN: Retrospective cohort design. OBJECTIVES: To compare socio-demographic, impairment characteristics and utilization (physician and emergency department (ED) visits) for non-traumatic (NTSCI) and traumatic (TSCI) spinal cord injury 1 year post inpatient rehabilitation. SETTING: Ontario, Canada. METHODS: Inpatient stays (2003-2006) were identified from the National Rehabilitation Registry System. Exclusions were: in-hospital mortality; discharge after 31 March 2006; death within 1 year after discharge. Multivariate logistic regression analyses were used to determine factors predicting high utilization. RESULTS: NTSCI cases (n=1002) were greater than TSCI (n=560). NTSCIs were older (mean=61.6, s.d.=15.8) with more co-morbidities, paraplegic (39.5%) and female (P<0.001). NTSCI had higher FIM admission and discharge scores but lower change scores. Mean number of physician visits for NTSCI and TSCI were 31.2 (median=24) and 29.7 (median=22), with no significant differences in mean specialist visits (NTSCI 16.5: TSCI 17.0). Factors predicting 30 or more physician visits included age 60 years or above (OR=1.5; 95% CI=1.2-1.9), urban living (OR=1.59; 95% CI=1.12-2.22) and lowest quartile (18-88) discharge FIM (OR=1.8; 95% CI=1.4-2.3). Charlson score of 3 or more (OR=2.1; 95% CI=1.3-3.2), urban living (OR=1.92; 95% CI=1.3-2.86) and lowest quartile discharge FIM (OR=1.5; 95% CI=1.2-2.0) were associated with 20 or more specialist visits. Factors for high ED use were: rurality (OR=1.5; 95% CI=1.1-2.1), low income (OR=1.4; 95% CI=1.1-1.9) and low (18-88) discharge FIM (OR=1.7; 95% CI=1.3-2.2). CONCLUSION: Both demonstrated significant health care utilization requiring attention to health care needs; particularly for those living in rural settings, with low income and/or low functional ability.

Health system factors associated with rehospitalizations after traumatic spinal cord injury: a population-based study.

STUDY DESIGN: This is a cohort study with 1-year follow-up. OBJECTIVES: The aim of this study was to examine 1-year rehospitalization rates following spinal cord injury (SCI) onset and health system factors affecting rehospitalization. METHODS: All persons who had an acute care hospitalization for traumatic SCI in Ontario between 1 April 2003 and 31 March 2006 were identified according to International Classification of Diseases, Tenth Revision codes and followed for 1 year following acute care discharge through record linkage of administrative databases. Index cases with an SCI admission the year before 2003 as well as persons who died within 1 year after the index hospitalization were excluded from the analysis. Factors associated with 1-year rehospitalization were assessed using multivariate logistic regression analyses and included age, sex, rurality, length of stay, comorbidity, level of injury, discharge disposition, in-hospital complication, physician visits and specialist visits measure and etiology of injury. RESULTS: A total of 559 individuals met the inclusion criteria and 27.5% (n=154) were rehospitalized 1 year after initial acute care discharge. Factors significantly associated with 1-year rehospitalization were length of stay, rural residence, 50+ outpatient physician visits and 50+ specialists visits following the index admission. The main causes of rehospitalization were musculoskeletal, respiratory, gastrointestinal and urological disorders. CONCLUSION: This study presents recent data on rehospitalization and yet rehospitalization rates continue to remain high. Our findings have significant implications for healthcare policy and planning in Ontario, Canada with respect to the management of SCI to achieve optimal health outcomes, in particular in rural areas.

Physician utilization among adults with traumatic spinal cord injury in Ontario: a population-based study.

STUDY DESIGN: Retrospective cohort study. OBJECTIVES: To describe the physician utilization patterns (family physicians (FP), specialist and emergency department (ED) visits) of adults with traumatic spinal cord injury (SCI) 1 year after the initial injury. SETTING: Ontario, Canada. METHODS: A total of 559 individuals with a traumatic SCI were identified. Five administrative databases were linked to examine health-care utilization in acute care, inpatient rehabilitation, chronic care rehabilitation, outpatient physician visits and ED visits. Factors predicting frequent physician, specialist and ED use were identified. RESULTS: The mean number of physician visits for traumatic SCI patients during the first year after their injury onset was 31.7 (median 26). FPs had the greatest number of visits (mean 11.6, median 7) followed by physiatrists (mean 6.1, median 2). Factors predicting 50 or more physician visits included age 70 or above (OR=3.6, 95% CI=2.0-6.5), direct discharge to chronic care (OR=3.6, 95% CI=1.0-13.1) and in-hospital complication (OR=2.34, 95% CI=1.3-4.3). Age 70 or less (OR=0.19, 95% CI=0.0-0.9) and direct discharge to chronic care were associated with 50 or more specialist visits. Only rurality predicted two or more visits to the ED. CONCLUSIONS: Individuals with traumatic SCI show significant physician utilization, especially among their FPs and physiatrists. Although the factors predicting higher physician and specialist utilization may reflect individuals with the most severe impairment, comorbid conditions or lack of social support, the model for higher ED visits may point to limited accessibility to/availability of primary care services for SCI patients in rural regions.

Geographical variations in hip fracture risk for women: strong effects hidden in standardised ratios.

UNLABELLED: There was 75% variation in the trend in hip fracture incidence with age in women aged 50 to 85 in France. In southwest France, the women are at higher risk of hip fracture at younger ages. This finding should be taken into account when examining risk factors. INTRODUCTION: Few studies have analysed the geographical variations in the relationship between age and hip fracture incidence. Our goal was to assess these variations among women under 85 within the same country. METHODS: The study population included women aged 50 to 85 who were living in France in 2004. Hip fracture cases were identified in the French Diagnosis Related Groups (DRG)-like database using the diagnosis code for closed hip fractures and procedural codes for treatment. The Moran index and a spatial model using latitude and longitude were used to assess the geographical heterogeneities of cumulative incidence risk (CIR) and age effect. RESULTS: A total of 29,218 hip fracture cases were identified. A south-to-north CIR gradient ranging from 7 to 16% was observed. The variation in the number of years until double hip fracture incidence was 75% (i.e. 1.49 to 2.57 years). In the south, and more markedly in southwest France, the women are at higher risk of hip fracture at a younger age. CONCLUSION: The risk of fracture may be different between women of the same age. This may be hidden in a comparison of standardised ratios. This finding should be considered when examining risk factors and implementing public health interventions.

Is it possible to estimate the incidence of breast cancer from medico-administrative databases?

One approach to estimate cancer incidence in the French Départements is to quantify the relationship between data in cancer registries and data obtained from the PMSI (Programme de Médicalisation des Systèmes d'Information Médicale). This relationship may then be used in Départements without registries to infer the incidence from local PMSI data. We present here some methodological solutions to apply this approach. Data on invasive breast cancer for 2002 were obtained from 12 Départemental registries. The number of hospital stays was obtained from the National PMSI using two different algorithms based on the main diagnosis only (Algorithm 1) or on that diagnosis associated to a mention of "resection" (Algorithm 2). Considering registry data as gold standard, a calibration approach was used to model the ratio of the number of hospital stays to the number of incident cases. In Départements with registries, validation of the predictions was done through cross-validation. In Départements without registries, validation was done through a study of homogeneity of the mean number of hospital stays per patient. Cross-validation showed that the estimates predicted by the model were true with data extracted by Algorithm 1 but not by Algorithm 2. However, with Algorithm 1, there was an important heterogeneity between French Départements as to the mean number of hospital stays per patient, which had an important impact on the estimations. In the near future, the method will allow using medico-administrative data (after calibration with registry data) to estimate Départemental incidence of selected cancers.

Trends in HRT and anti-osteoporosis medication prescribing in a European population after the WHI study.

UNLABELLED: To assess the prescription patterns of anti-osteoporosis medications, three cross-sectional analyses were performed between 2004 and 2006. Women aged 50 and older were identified from the health insurance claims database of the Rhône-Alpes area. HRT prescriptions decreased while bisphosphonates and raloxifene prescriptions increased, respectively, in different age groups. INTRODUCTION: The objective of this study was to assess the prescription patterns of hormone replacement therapy (HRT) and anti-osteoporosis medications (AOM) in post-menopausal French women since the WHI and the revision of the French clinical practice guidelines in 2004. METHODS: Three cross-sectional analyses were performed between 2004 and 2006. Women aged 50 and older who had at least one claim for a prescription for HRT, bisphosphonates or raloxifene were identified from health insurance claims database of the Rhône-Alpes area. RESULTS: A 39% decrease in the number of women who had HRT was observed (67,241 to 41,024). Twenty-one percent and 18% increases were observed, respectively, for bisphosphonates (39,192 to 47,395) and raloxifene (10,263 to 12,060). HRT and raloxifene were mainly prescribed to women aged 55 to 64 (58% and 39%, respectively), bisphosphonates to women aged 65 to 84 (70%). Ninety-eight percent of women had HRT prescribed by a gynaecologist or a general practitioner (GP). Most AOM were prescribed by a GP; 13% of women had AOM prescribed by a rheumatologist. CONCLUSION: Prescriptions for HRT in post-menopausal French women have significantly decreased while bisphosphonates and raloxifene prescriptions have increased, respectively, in different age groups but to a lesser extent than the HRT decrease.

Nine-year downward trends in surgical site infection rate in southeast France (1995-2003).

The aim of this study was to estimate temporal trends in the incidence of surgical site infection (SSI) using a large SSI surveillance network in southeast France from 1995 to 2003. Data were analysed from 187 surgical wards that had participated in the network for at least two years. The change in SSI rate over time was modelled using a hierarchical logistic regression model with patients clustered within surgical wards. Of the 200 207 patients selected, 3786 (1.9%) had an SSI. The nine-year trend in SSI rate estimated by an odds ratio of 0.95 (95% confidence interval 0.93-0.97) was interpreted as a 5% decrease in SSI rate per year. This decrease was constant over the study period and was observed for almost all of the different types of surgical operations (orthopaedic, gastrointestinal, urology, etc). Overall SSI rates were reduced by 45% over a period of nine years. This trend was maintained even when taking into account the heterogeneity of the surgical wards and the diversity of patient demographics over time. From this, the 5% decrease per year can be reasonably interpreted as a result of preventive measures taken by surgical wards to reduce SSIs.

[Quality and usefulness of an anonymous unique personal identifier to link hospital stays recorded in French claims databases]. / Qualité et utilité d'un identifiant patient anonyme et unique pour le chaînage des séjours hospitaliers dans les bases de données médicoéconomiques françaises.

BACKGROUND: Since 2001, the French national case mix program is allowed by law to use an enciphering algorithm named "FOIN" to produce a unique anonymous identifier in order to crosslink, within and across hospitals, discharge abstracts from a given patient. This algorithm "thrashes" the person's health insurance number, date of birth and gender. Before using information produced by the case mix program, either for case mix payment or for epidemiology research or for assessing care approaches, the quality of linkage must be evaluated. METHODS: Foin error flags were first assessed in the 2002 Rhône-Alpes regional case mix database. Second, for the two university hospitals of Lyon and Saint-Etienne, double identifiers (two or more Foin identifiers for the same patient) and collisions (a single Foin identifier for at least two patients) were compared with others identifiers: administrative identifier and an anonymous identifier produced by Anonymat software from name, forename and date of birth. Third, Foin error flags are crossed with Foin double identifier or collision mistakes. RESULTS: First, among 1,668,971 hospital discharge abstracts from the regional case mix database, 206,710 (12.4%) had at least one Foin error flag. The most frequent error flag (93026 [5.5%] stays) was due to the lack of the three identifying variables. The greatest number for error flags concerned the stays of newborns (38.5%) and those of public hospitals (17.3%). Second, Foin created a few double identifiers: 1.2% among 137,236 patients from university hospital of Lyon and 0.3% among 39512 patients from university hospital of Saint-Etienne. The collisions concerned 7776 (5.7%) patients from Lyon and 460 (1.2%) from Saint-Etienne. The identifier produced by Anonymat performed better than the one produced by Foin: 99.6% from the two university hospitals. Third, less than 3% of stays without Foin error flag nevertheless had mistakes on Foin when compared with others identifiers. CONCLUSION: The overall assessment is not in favour of a quality threshold using the Foin identifier on a routine basis except in some areas and if certain activities like neonatology are excluded. There are several ways to improve the linkage of health data.

[Role of hospital claims databases from care units for estimating thyroid cancer incidence in the Rhône-Alpes region of France]. / Place du Programme de médicalisation des systèmes d'information (PMSI) dans l'estimation de l'incidence du cancer de la thyroïde dans la région Rhône-Alpes.

BACKGROUND: Incidence measures are essentially based on the data collected by cancer registries. Hospital claims databases from care units (PMSI) can be used as a source of information for registries because they contain standard records of most cancer patients. Regarding thyroid cancer, we have evaluated the PMSI as a source of information for the Rhône-Alpes thyroid cancer registry and usefulness of PMSI as a tool for surveillance of thyroid cancer incidence. METHODS: Patients with incident thyroid cancer in 2002 were identified in the claims data of the Rhône-Alpes region using an algorithm based on DRG codes of thyroidectomy and on diagnosis codes of thyroid cancer in a principal or secondary position. The patients identified were compared to those in thyroid cancer registry of the Rhône-Alpes region regarding sex, age, ZIP code of residence, month of discharge and length of stay versus the diagnosis date. When the percentage of cases of claims data identified in the cancer registry and the percentage of cases of the cancer registry identified in claims data were obtained, the capture-recapture method was applied to estimate the number of missing cases and the total number of incident thyroid cancers in the region. RESULTS: 667 patients were identified in claims data while the cancer registry included 677 patients. 95.2% of patients identified in claims data were in the cancer registry and 82.3% of patients in the cancer registry were identified in claims data. Cases lacking in claims data mostly corresponded to micro-cancers which represented 41% of cases in the cancer registry. Regarding cancer above 1 cm, 92% of the cancer registry cases were identified in claims data. Sensitivity of combining information from cancer registry and claims data was 99.2%. Cases lacking in cancer registry, present in claims data base and considered as true cases after obtaining pathological confirmation represented 2% of the whole thyroid cancer population. CONCLUSION: Claims data obtained from anonymous regional or national bases can be helpful for checking the completeness of thyroid cancer registries and to provide a small amount of unknown cases. They can be considered an acceptable tool for surveillance of thyroid cancer incidence. The significance of the variations in incidence that could be observed from claims data remains to be evaluated in comparison with comparable data obtained from registries.

Estimates of the number of cancer patients hospitalized in a geographic area using claims data without a unique personal identifier.

OBJECTIVE: In French national claims databases, claims are currently anonymous i.e. not linked to individual patients. In order to improve our estimate of the medical activity related to cancer in one French region, a statistical method was developed to use claims data to assess the number of cancer patients hospitalized in acute care. METHODS: This method used the medical and administrative information available in the claims (i.e. age, primary site, length of stay) to predict an average number of stays per patient, followed by a number of patients. It was based on a two-phase study design using an internal dataset which contained personal identifiers to estimate the model parameters. RESULTS: The predicted number of acute care patients hospitalized in one or several health care centers in one French region was 38,109 with a 95% predictive interval (37,990; 38,228) for the first six months of 2002. A prediction error of 24 per thousand was found. CONCLUSION: We provide a good estimate of the morbidity in acute care hospitals using claims data that is not linked to individual patients. This estimate reflects the medical activity and can be used to anticipate acute care needs.

[Sensitivity and specificity of two methods used to identify incident breast cancer in specialized units using claims databases]. / Sensibilité et spécificité de deux méthodes d'identification des cancers du sein incidents dans les services spécialisés à partir des données médico-administratives.

BACKGROUND: Hospital claims databases from acute care units are available nationwide and contain most patients at the beginning of their cancer. The goal is to define the ability of these databases to provide a number of incident breast cancer cases using identification methods. Two identification methods were assessed in three specialized sections of a teaching hospital. METHODS: The first method identified women who had at least one stay with a principal diagnosis of breast cancer. The second, which is more restrictive, identified women who had at least one stay with a principal diagnosis of breast cancer and a breast cancer-specific surgical treatment code. Both methods were applied to 4588 women 20 Years of age or older hospitalized in three specialized sections of the Hospices Civils de Lyon in 2000. To categorize these women in two groups, incident breast cancer cases or non-incident breast cancer cases, 150 women were randomized in each of two groups, one for incident breast cancer cases and one for non-incident breast cancer cases. Their medical records were used as references. RESULTS: Sensitivity, specificity and their credibility intervals were respectively 99.4% (84-99.9) and 91.7% (90.3-93.3) for the first method and 93.8% (76.2-98.7) and 97.3% (96.1-98) for the second. Among women wrongly identified with an incident breast cancer in 2000, 75.4% (43/57) had a breast cancer that was not incident that Year with the first method, compared to 96% (24/25) with the second. Among these women wrongly identified with an incident breast cancer, coding errors of the principal diagnosis were found for 24.6% (14/57) of patients with the first method and for 4% (1/25) with the second. Their correction led to 99.2% (86.5-99.9) sensitivity and 92.9% (91.4-94.6) specificity for the first method and to 94.2% (76.5-98.7) sensitivity and 97.3% (96.2-98.1) specificity for the second. CONCLUSIONS: The second method using cancer-specific surgical codes appeared more specific with a slight loss in sensitivity. The use of identification methods to assess the number of incident cancer cases still have to be defined.

Two-phase study to assess the number of cases based on claims databases: characteristics of the validation data set.

OBJECTIVE: In a two-phase study design, the characteristics of an external data set were studied for precision and bias of the number of incident or prevalent cases of a disease obtained from claims databases. METHODS: In the study population (first phase), incident or prevalent cases were counted whereas external data (second phase) provided sensitivity and specificity estimates to count cases in a claims database. Influence of potential differences in sensitivity and specificity between the two phases were evaluated. This was illustrated for 50-90% sensitivity and 99-99.99% specificity ranges. RESULTS AND CONCLUSIONS: The impact of differences in sensitivity and specificity depends on the odds of disease in the study population. We provide advice on the choice of adequate data sets to correct claims database estimates.

Method of correction to assess the number of hospitalized incident breast cancer cases based on claims databases.

Estimations of the number of hospitalized incident cancer cases show biases when claims databases are used. This is due to false reports of incident cancer because of a lack of specificity, and because of unrecorded cancers resulting from a lack of sensitivity. We present a statistical method to provide corrected estimations. This method is based on a two-phase study design using an external data set for sensitivity and specificity estimates. Inaccuracy of the corrected number of hospitalized incident cancer cases was assessed by a credibility interval determined by a Bayesian approach using a Monte Carlo method. Based on the population hospitalized in a large group of French University hospitals, 334 women were identified in the French claims database as having potential incident cases of breast cancer in 1997. According to our method, the corrected number was 565 (550-580). In absence of hospital-based cancer registries, our approach provides estimates and credibility intervals, and has many potential applications in defining hospital policies with its applicability to other diseases.