Validity of hospital discharge diagnoses for the assessment of the prevalence and incidence of amyotrophic lateral sclerosis.

INTRODUCTION: As amyotrophic lateral sclerosis (ALS) is a rare neurological disorder, many sources must be investigated for complete case ascertainment. Hospital discharge diagnoses (HDDs) are a source of administrative data for patients hospitalized for a given disease. The accuracy of HDDs was tested, to assess whether they could be used to trace patients with amyotrophic lateral sclerosis (ALS). METHODS: All the HDDs referring to patients with motor neuron disease (MND) (ICD-9 code 335.2) were identified from primary hospital discharge diagnoses made during the years 1994-95 in six target provinces of the Lombardy region in Northern Italy (total population 2 922 920). The medical records of the patients were reviewed and each patient was then accepted or rejected according to whether he/she did or did not satisfy the El Escorial criteria for the diagnosis of ALS. The ALS discharge diagnoses filed in the study period in the neurology departments of all the regional hospitals were the 'gold standard'. Newly diagnosed patients registered during the year 1998 in the same provinces served for the calculation of the incidence of the disease. Sensitivity, specificity and positive predictive value were calculated. Data were presented separately for all the patients and those who were indicated as first hospital admissions. RESULTS: During the study period a total of 1 012 122 hospital discharge records were coded in the six index provinces. The ICD-9 code 335.2 was reported in 408 of these (0.04%). These discharge diagnoses were made in 267 patients, 245 of whom were resident in Lombardy at the time of hospital admission. The hospital records of 237 cases were examined. In 95 of these the medical records showed a diagnosis other than ALS. Other motor neuron diseases were listed for 50 patients. Thirteen additional cases were traced through neurology department files. The sensitivity of the HDDs was 91.6%, the specificity 99.9%, the positive predictive value 65.4%. The corresponding values for first-hospitalized patients were 97.7, 99.9, and 37.3. The overall ALS incidence rate obtained from the HDDs was 2.1 per 100 000 (95% CI 1.3-3.3). The corresponding rate calculated after exclusion of the false positives and inclusion of the false negatives was 0.8 (95% CI 0.1-2.1) and that obtained using the 1998 cases registered in the same areas was 1.6 (95% CI 0.7-2.7). CONCLUSIONS: The incidence of ALS is overestimated when HDDs are used for the estimate. Nevertheless, with improved quality, administrative data could be useful for the estimate of prevalence and incidence. [ABSTRACT FROM AUTHOR]