[Predictive value and sensibility of hospital discharge system (PMSI) compared to cancer registries for thyroïd cancer (1999-2000)]

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.

A novel approach for medical research on lymphomas: A study validation of claims-based algorithms to identify incident cases

The use of claims database to study lymphomas in real-life conditions is a crucial issue in the future. In this way, it is essential to develop validated algorithms for the identification of lymphomas in these databases. The aim of this study was to assess the validity of diagnosis codes in the French health insurance database to identify incident cases of lymphomas according to results of a regional cancer registry, as the gold standard.Between 2010 and 2013, incident lymphomas were identified in hospital data through 2 algorithms of selection. The results of the identification process and characteristics of incident lymphomas cases were compared with data from the Tarn Cancer Registry. Each algorithm's performance was assessed by estimating sensitivity, predictive positive value, specificity (SPE), and negative predictive value.During the period, the registry recorded 476 incident cases of lymphomas, of which 52 were Hodgkin lymphomas and 424 non-Hodgkin lymphomas. For corresponding area and period, algorithm 1 provides a number of incident cases close to the Registry, whereas algorithm 2 overestimated the number of incident cases by approximately 30%. Both algorithms were highly specific (SPE = 99.9%) but moderately sensitive. The comparative analysis illustrates that similar distribution and characteristics are observed in both sources.Given these findings, the use of claims database can be consider as a pertinent and powerful tool to conduct medico-economic or pharmacoepidemiological studies in lymphomas.

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.

[Role of French hospital claims databases from care units in epidemiological studies: the example of the "Cohorte Enfant Scanner" study]

BACKGROUND

The "Cohorte Enfant Scanner", a study designed to investigate the risk of radiation-induced cancer after childhood exposure to CT (computed tomography) examinations, used clinical information contained in the "programme de médicalisation des systèmes d'information" (PMSI) database, the French hospital activities national program based upon diagnosis related groups (DRG). However, the quality and adequacy of the data for the specific needs of the study should be verified. The aim of our work was to estimate the percentage of the cohort's children identified in the PMSI database and to develop an algorithm to individualize the children with a cancer or a disease at risk of cancer from medical diagnoses provided by the DRGs database.

METHODS

Of the 1519 children from the "Cohorte Enfant Scanner", who had had a CT scan in the radiology department of a university hospital in 2002, a cross linkage was performed with the DRGs database. All hospitalizations over the period 2002-2009 were taken into account. An algorithm was constructed for the items "cancer" and "disease at risk for cancer" on a sample of 150 children. The algorithm was then tested on the entire population.

RESULTS

Overall, 74% of our population was identified in the DRGs database. The algorithm individualized cancer diagnoses with 91% sensitivity (95% confidence interval [95%CI]: 86%; 97%) and 98% specificity (95%CI: 97%; 99%) and 86% positive predictive value (95%CI: 80%; 93%). For the diagnosis of disease at risk for cancer, the sensitivity, specificity and positive predictive value were respectively 91% (95%CI: 84%; 98%), 94% (95%CI: 92%; 95%) and 52% (95%CI: 43%; 61%).

CONCLUSION

The DRG database identified with excellent sensitivity and specificity children with diagnoses of cancer or disease at risk for cancer. Hence, potential confounding factors related to the disease of the child can be taken into account for analyses performed with the cohort.

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.

[Performance evaluation of hospital claims database for the identification of incident central nervous system tumors compared with a cancer registry in Gironde, France, 2004]

BACKGROUND

Cancer registries cover 18% of the French population. A national surveillance might be warranted for some potentially environment-related cancers such as tumors of the central nervous system (CNS) to detect abnormal incidence variations. The PMSI database provides an interesting source of comprehensive, standardized and mandatory data collected from all health facilities. The aim of this work was to develop methods to identify incident CNS tumors using the PMSI database.

METHODS

A selection of patients living in Gironde was made in the 2004 PMSI database of the hospital of Bordeaux, using the CNS tumors codification. Cases were validated via the CNS primary tumor registry of Gironde taken as the reference, or medical records. Various combinations of criteria were defined and tested.

RESULTS

The first selection based on diagnoses identified patients with a sensitivity of 84% and a positive predictive value (PPV) of 34%. Patients wrongly identified by the PMSI were non-incident cases (49%) or patients without a CNS tumor (45%). Patients with a tumor not identified by the PMSI had been hospitalized in 2005 (44%) or had no code for CNS tumor (42%). According to the algorithms, the sensitivity ranged from 64% to 84%, and the PPV from 34% to 69%. The best combination had a sensitivity of 67% and a PPV of 69% and was obtained with codes for CNS tumor in 2004 associated with a diagnostic or therapeutic code for persons under 70 years without code for CNS tumor in previous years or code for metastasis in 2004.

CONCLUSION

According to these results, the PMSI database cannot be used alone to calculate the incidence of these complex tumors. However the PMSI database plays an important role in cancer surveillance, in combination with other information sources and the expertise of cancer registries. This role could increase with further reflection and improvement of data quality.

Increasing incidence of zygomycosis (mucormycosis), France, 1997-2006

Results were derived from a population-based study using hospital discharge data.

We analyzed hospital records to provide a population-based estimate of zygomycosis incidence and trends over a 10-year period at a national level in France. Data showed an increasing incidence from 0.7/million in 1997 to 1.2/million in 2006 (p<0.001). We compared our data with those from the French Mycosis Study Group, a recently established voluntary network of French mycologists coordinated by the National Reference Center for Mycoses and Antifungals. We documented that incidence of zygomycosis increased, particularly in patients with hematologic malignancies or bone marrow transplants. The role of previous exposure to antifungal drugs lacking activity against zygomycetes could explain this increase but does not appear exclusive. Incidence also increased in the population of patients with diabetes mellitus. We conclude that observed trends reflect a genuine increase of zygomycosis cases in at-risk populations.

[Estimating the incidence of colorectal cancer in France from a hospital discharge database, 1999-2003]

BACKGROUND

Cancer incidence in France is monitored by district-level cancer registries, which cover only 15% of the population. Incidence at the national and regional level are estimated from mortality data by extrapolating the ratio between incidence and mortality observed in the districts covered by a cancer registry. Using the incidence/mortality ratio might not be relevant at the district-level (département). This study aims to produce district-level estimations of colorectal cancer incidence, using the ratio between incident cases from cancer registries and surgery admissions for colorectal cancer identified in the national hospital discharge database.

METHODS

This ratio was studied for the period 1999-2003 in the 13 districts covered by a cancer registry. For each sex separately, the number of incident cases was analyzed according to the number of surgery admissions for resection of colorectal cancer using a Poisson model. Age was introduced in the model as a fixed effect and district as a random effect. The model's ability to predict incidence was tested through cross-validation. The model was then extrapolated in order to estimate incidence of colorectal cancer in all French districts.

RESULTS

In the districts covered by a cancer registry, cross-validation showed the model had a good predictive ability, except in men for one district where the difference between predicted and observed incidence reached 10%. Estimated incidence rates, age-standardized on the world population, ranged broadly from 29 to 44 per 100,000 in men and from 17 to 27 per 100,000 in women. Incidence did not show any clear geographical pattern.

CONCLUSION

Among districts covered by a cancer registry, cross-validation showed overall good accuracy of predicted incidence. Inclusion of several admissions per patient was certainly a minor source of error in these estimations. Indeed, our selection only included 2% of multiple admissions, without geographical variations, in 2002 and 2003, years for which patient identifiers were available in the hospital discharge database. Estimated incidence rates presented moderate geographical variations and their prediction intervals should be taken into account.

[Follow-up of cancer treatment activities at the University Teaching Hospital in Dijon: the value of data from standardized discharge summaries]

This work aimed to assess the feasibility of coalescing data sources to follow up on the management and delivery of cancer patient care using both anatomical-pathological data and a medical record information system based on the "Medicalisation of Information Systems Programme" which is administered by an independent source. The study group was comprised of all hospitalised cancer patients in the Dijon university teaching hospital during the first quarter of 2001. Data were obtained from a cross-analysis of medical records with pathological information and standard discharge files. A manual validation was then carried out to ensure a valid synthesis. Overall, 1377 abstracts were created and selected for cancer patients hospitalized for the first time. Among these, 60% were validated by the compatibility and concordance of data between the medical record discharge issued and the pathological record, less than 5% were identified only through looking at the anatomical-pathological data in the medical record, and 24% identified by exploring the patient discharge forms. These results demonstrate the difficulty and challenge as well as the benefits of crossing multiple sources of patient information and combine them to more thoroughly and appropriately assess the hospital's activity for caring for cancer patients.

Estimating infra-national and national thyroid cancer incidence in France from cancer registries data and national hospital discharge database

OBJECTIVE

As in many countries, cancer registries cover only part of the population in France. Incidence/mortality ratio observed in registries is usually extrapolated to produce national estimates of cancer incidence. District-level estimates are not currently available. For cancer sites such as thyroid, the incidence/mortality ratio widely varies between districts, and alternative indicators must be explored. This study aims to produce national and district-level estimations of thyroid cancer incidence in France, using the ratio between incidence and hospital-based incidence.

METHODS

Analyses concerned population living in France and aged over 20, for the period 1998-2000. For each sex, number of incident cases were analysed according to number of surgery admissions for thyroid cancer (Poisson model) in the districts covered by a registry. Age was included in the model as fixed effect and district as random effect. The model's ability to predict incidence was tested through cross-validation. The model was then extrapolated to produce national incidence estimations, and for women, district-level estimations.

RESULTS

The national estimations of incidence rate age-standardised on the world population were 3.1 [95% prediction interval: 2.8-3.4] for men and 10.6 [9.8-11.4] for women, corresponding respectively to 1,148 [1,042-1,264] and 4,104 [3,817-4,413] annual new cases. For women, district-level incidence rates presented wide geographical variations, ranging broadly from 5 to 20 per 100,000. These estimations were quite imprecise, but their imprecision was smaller than the extent of geographical disparities.

CONCLUSION

National incidence estimations obtained are relatively precise. District-level estimations in women are imprecise and should be treated carefully. They are informative though regarding the extent of geographical disparities. The approach can be useful to improve national incidence estimates and to produce district-level estimates for cancer sites presenting a high variability of the incidence/mortality ratio.