2.1 Eleven nursing homes managed by Beverly Enterprises Inc. were selected; one as a pilot site in which to test protocols for data collection and study procedures and ten for the study. A complete list of elderly residents (65 years of age or older) compiled by nursing home staff, contained the nursing home record number, date of birth, date of admission, nursing home identified epilepsy or seizure disorder (yes or no) and current AED use (yes or no) of each elderly resident. Records for review were selected from each nursing home's list in three steps:
- Records for all elderly residents with nursing home identified epilepsy or seizure disorders were selected.
- Records of non-epilepsy/seizure disorder residents were matched, on a one-to-one basis, with the epilepsy/seizure disorder residents by age group (65 to 74, 75 to 84, 85 and over), gender and closest first admission date. Records for these matched residents without epilepsy or seizure disorder were selected without regard to antiepileptic drug use.
- Records for all remaining residents without a diagnosis of seizures or epilepsy who received AEDs were selected so that abstractors could not assume a resident's diagnostic status by their treatment.
2.2 At each nursing home, abstractors searched selected residents' records for the most recent full MDS, either the last Annual MDS assessment or the MDS completed at admission. Epilepsy or seizure information was collected from section I, Items 1 and 3 on the paper MDS contained in the most recent MDS assessment ().
Data Items Abstracted for Validation of Documentation of Epilepsy or Seizure Disorder.
The study neurologist (IEL) conducted chart review sessions at each nursing home. The study neurologist was masked from knowledge of the resident's epilepsy or seizure disorder classification on the MDS. He then searched the remainder of each nursing home record for any mention of an epilepsy or seizure diagnosis and any supporting information. This included any consultations by a neurologist and results of MRIs and EEGs if available in the nursing home record. In many cases, a discharge summary from a hospital was also available for review.
2.3 Approval to collect these data was obtained from the University of Minnesota's International Review Board and Beverly Enterprises, Inc.
2.4 From the paper MDS, a resident was considered to have a seizure disorder or epilepsy if:
- The “Seizure Disorder” item (MDS Section 1. Disease Diagnoses, Item 1. aa.) was checked or
- ICD-9 codes 345.XX or 780.3X (MDS Section I. Disease Diagnoses, Item 3.a-e) were listed or
- The words “epilepsy,” “seizure” or “convulsion” were listed in the text section (MDS Section I. Disease Diagnoses, Item 3. a-e).
These same criteria were used for the computerized MDS. However, text documenting epilepsy or seizure disorder was not available in the computerized MDS file. From the neurologist's data collection session, the resident was considered to have epilepsy or a seizure if the neurologist saw a physician's note using these terms, a nurses note describing a seizure, a hospital discharge summary listing these terms, and any results of diagnostic tests or neurologist's consultations supporting use of these terms. The neurologist did not attempt make a diagnosis independently, rather limited the review to use of these terms by the physicians providing care for these patients.
2.5 Percent agreement between the study neurologist's determination of epilepsy or seizures from the nursing home record and documentation of epilepsy or seizure disorder from the “paper MDS” in the nursing home record was calculated using the study neurologist's determination as the “gold standard”. Percent agreement was also calculated between epilepsy or seizure disorder on the computerized MDS and epilepsy or seizure disorder on the “paper MDS” abstracted in nursing homes.
Statistical analysis was completed using SPSS® for Windows® software. Percent agreement, positive and negative predictive values, and the kappa statistic were computed for both validity comparisons (Streiner DL & Norman GR 2003
). Agreement (yes/no) was compared across age groups (65-74,75-84,85+) and between genders using Chi Square tests. Length-of-stay (time since first admission) and time from MDS to data abstraction were compared between agreement groups (yes/no) using independent t-tests.