Elsevier

Epilepsy & Behavior

Volume 51, October 2015, Pages 65-72
Epilepsy & Behavior

Development and validation of an epidemiologic case definition of epilepsy for use with routinely collected Australian health data

https://doi.org/10.1016/j.yebeh.2015.06.031Get rights and content

Highlights

  • Reasonably accurate epilepsy case ascertainment from unpublished Australian data

  • Confirms similar findings to other international datasets

  • ICD-10AM codes (G40 and G41) and antiepileptic medications conferred the highest PPV.

  • National hospital datasets in Australia are limited to hospital inpatients.

  • Noncapture of ED and hospital outpatients has methodological issues.

Abstract

Objectives

We report the diagnostic validity of a selection algorithm for identifying epilepsy cases.

Study design and setting

Retrospective validation study of International Classification of Diseases 10th Revision Australian Modification (ICD-10AM)-coded hospital records and pharmaceutical data sampled from 300 consecutive potential epilepsy-coded cases and 300 randomly chosen cases without epilepsy from 3/7/2012 to 10/7/2013. Two epilepsy specialists independently validated the diagnosis of epilepsy. A multivariable logistic regression model was fitted to identify the optimum coding algorithm for epilepsy and was internally validated.

Results

One hundred fifty-eight out of three hundred (52.6%) epilepsy-coded records and 0/300 (0%) nonepilepsy records were confirmed to have epilepsy. The kappa for interrater agreement was 0.89 (95% CI = 0.81–0.97). The model utilizing epilepsy (G40), status epilepticus (G41) and ≥ 1 antiepileptic drug (AED) conferred the highest positive predictive value of 81.4% (95% CI = 73.1–87.9) and a specificity of 99.9% (95% CI = 99.9–100.0). The area under the receiver operating curve was 0.90 (95% CI = 0.88–0.93).

Conclusion

When combined with pharmaceutical data, the precision of case identification for epilepsy data linkage design was considerably improved and could provide considerable potential for efficient and reasonably accurate case ascertainment in epidemiological studies.

Introduction

Data linkage is an emerging powerful tool, particularly for ascertaining low incidence events, enabling medical diseases and health outcomes to be connected using routinely collected centralized databases. For the study of epilepsy, it has the advantage of efficiently identifying large samples of patients with epilepsy [1], [2], [3]; however, misclassification of cases with epilepsy in administrative databases is a major issue, limiting its utility as an epidemiologic instrument for disease surveillance and research.

International Classification of Diseases (ICD)-coded data are utilized internationally as an epidemiological tool for collection of national health statistics [4]; however, their use requires validation. Previous studies that utilize this classification system have used ICD-9 and ICD-10 versions [5], [6], [7], [8], [9], [10]. As ICD-10 revisions are country-specific and coding practices may vary between regions, it has been recommended that specificity and predictive values be evaluated for each population studied [11]. Current epidemiological guidelines suggest that a probable diagnosis of epilepsy can be made if 1 of the following 3 conditions is met: one medical encounter with a 3-digit code of G40.x (epilepsy); ≥ 2 medical encounters on separate days coded with G41 (status epilepticus) or with a 4-digit code R56.8 (other and unspecified convulsions); and a single medical encounter coded as other and unspecified convulsions (R56.8) and an antiepileptic drug prescription for three or more months [11]. A suspected diagnosis of epilepsy can be made with single episodes coded with R56.8 or G41 [11].

The coding of Australian Modification (AM) version of ICD-10 (ICD-10AM) clinical data in Australia exists for diagnoses and procedures of acute admitted patient episodes only [12]. At present, epilepsy coding for the ICD-10AM has not been validated. We sought to estimate the diagnostic accuracy of different ICD-10AM coding algorithms to identify patients with epilepsy in an Australian hospital setting and to develop an algorithm combining other routinely nationally collected data to maximize its precision for potential epidemiologic surveillance and research.

Section snippets

Methodology

Following a discharge from a public or a private hospital in Australia, all principal diagnoses and additional diagnoses (up to 99) in medical records are coded with ICD-10AM codes and submitted to the National Hospital Morbidity Database for statistical reporting [13], [14]. We retrospectively recruited a validation cohort by identifying cases with epilepsy and their common differential diagnoses in a metropolitan adult hospital setting with a large neurology unit including epilepsy

Results

During the study period, 42,760 patients were seen at our hospital in the ED and inpatient units. The records of 600 case episodes were reviewed, with 158 confirmed to have epilepsy (72/300 in the ED sample, and 86/300 in the inpatient sample, Χ2 (1) = 1.46, p < 0.23). This represents 52.3% of cases coded with epilepsy (G40), status epilepticus (G41), or ‘other and unspecified convulsions’ (R56.8) having confirmed epilepsy. There were 20 first unprovoked seizures and 5 incident epilepsy cases

Summary of findings

The linking of multiple sources of data has been recommended by the ILAE Commission on Epidemiology to improve epilepsy case ascertainment from administrative datasets [11]. This study, the first to be conducted in Australia, found that the identification of epilepsy in an adult hospital setting can be reasonably accurate and comparable with other countries of similar economic development [6], [7], [27] when based on a combination of data elements: ICD-10AM coding for epilepsy (G40), status

Conclusion

Our results demonstrated that the combination of administrative and prescription databases is necessary when designing privacy preserving data linkage studies in an Australian population. While this method provides considerable potential for future epilepsy epidemiological studies, limitations in the collection of Australian administrative data to hospital inpatient populations restrict its use. Incidence and prevalence studies are likely to be affected by selection bias given the lack of

Acknowledgments

We also acknowledge Ms. Esther Taylor from the data integrity and application support department at St Vincent's Hospital, Melbourne for her assistance in data acquisition.

Disclosure

Assoc. Prof. D'Souza has received travel, investigator-initiated, and speaker honoraria and has served on scientific advisory boards from UCB Pharma; has received educational grants from Novartis Pharmaceuticals, Pfizer Pharmaceuticals, and Sanofi-Synthelabo; has received educational, travel, and fellowship grants

References (48)

  • C.J. Wotton et al.

    Coexistence of schizophrenia and epilepsy: record-linkage studies

    Epilepsia

    (2012)
  • T. Khan et al.

    Epilepsy and the subsequent risk of cerebral tumour: record linkage retrospective cohort study

    J Neurol Neurosurg Psychiatry

    (2011)
  • C. Kwon et al.

    Motor vehicle accidents, suicides, and assaults in epilepsy: a population-based study

    Neurology

    (2011)
  • D.C. Bergen et al.

    Revising the ICD-10 codes for epilepsy and seizures

    Epilepsia

    (2012)
  • M.J.V. Pugh et al.

    The impact of epilepsy on health status among younger and older adults

    Epilepsia

    (2005)
  • N. Jetté et al.

    How accurate is ICD coding for epilepsy?

    Epilepsia

    (2010)
  • C. Franchi et al.

    Validation of healthcare administrative data for the diagnosis of epilepsy

    J Epidemiol Community Health

    (2013)
  • K. Tu et al.

    Assessing the validity of using administrative data to identify patients with epilepsy

    Epilepsia

    (2014)
  • D.J. Thurman et al.

    Standards for epidemiologic studies and surveillance of epilepsy

    Epilepsia

    (2011)
  • Metadata and Classifications Unit, Australian Institute of Health and Welfare. Coded clinical data. AihwGovAu n.d....
  • Australian Institute of Health and Welfare 2014

    Australian hospital statistics 2012–13

    (2014)
  • Metadata and Classifications Unit, Australian Institute of Health and Welfare

    Admitted patient care NMDS 2012–13 — long form exported from METeOR (AIHW's Metadata Online Registry)

    (2015)
  • D. Trewin

    Information paper: census of population and housing — socio-economic indexes for areas, Australia (2001)

    (2003)
  • E. Beghi et al.

    Recommendation for a definition of acute symptomatic seizure

    Epilepsia

    (2010)
  • Cited by (24)

    • Knowledge, attitudes, and practices of physiotherapists with regard to epilepsy and patients with epilepsy: A systematic scoping review

      2021, Epilepsy and Behavior
      Citation Excerpt :

      Epilepsy is one of the most common chronic neurological disorders of the brain that is characterized with unprovoked seizures [1,2].

    • Validating the accuracy of administrative healthcare data identifying epilepsy in deceased adults: A Scottish data linkage study

      2020, Epilepsy Research
      Citation Excerpt :

      Various potential sources of error remain possible, such as an incorrect diagnosis of epilepsy being listed by the death certifying doctor in a patient who did not actually have epilepsy (Death Certifcation Review Service, 2016), or human error occurring during the subsequent coding process (O’Malley et al., 2005). For example, the G40–41 ICD-10 codes for epilepsy and status epilepticus are often incorrectly coded as G43.1 (migraine) or G45.x (transient cerebral ischemia) (Jette et al., 2010; Tan et al., 2015). Furthermore, a deceased PWE might not be identified in an epilepsy mortality study if death certificates are the only source for case-ascertainment.

    • Accuracy of ICD-10-CM claims-based definitions for epilepsy and seizure type

      2020, Epilepsy Research
      Citation Excerpt :

      The prior International League Against Epilepsy (ILAE) classification (2014) posed several problems for classifying epilepsy and seizure type (Jette et al., 2015) in previous studies that examined the accuracy of ICD-10 codes provided by general practitioners or neurologists against the prior ILAE classification and terminology (Jette et al., 2010b; Reid et al., 2012; Tan et al., 2015; Tu et al., 2014). Petit mal seizure terminology of ICD-10-CM, for example, had no ILAE equivalent (Tan et al., 2015). We found that our ICD-10-CM claims-based definitions performed well in identifying patients with true focal seizures (PPV = 86.8 %; Sp = 79.4 %) and generalized seizure type (PPV = 84.9 %, Sp = 95.4 %); using these ICD-10-CM code groups for research in the outpatient epilepsy clinic setting might capture a high percentage of those with true seizure types.

    • Hospital service utilization trajectories of individuals living with epilepsy in New South Wales, Australia, 2012–2016: A population-based study

      2020, Epilepsy and Behavior
      Citation Excerpt :

      For instance, auditing has identified good-to-excellent classification of principal and comorbid diagnosis codes in routinely collected Australian hospital data generally [47]. A validation of epilepsy diagnosis codes in Australian hospital data indicated moderate to high positive predictive values for G40 and G41 cases (i.e., 70% and 90%, respectively), and the number of G40 and G41 cases needed to be coded to misdiagnose one case was 455 [48]. Consequently, the expected number of misclassified epilepsy cases in the present study is likely to be relatively low.

    • SUDEP and mortality in epilepsy: The role of routinely collected healthcare data, registries, and health inequalities

      2020, Epilepsy and Behavior
      Citation Excerpt :

      Generally though, disease diagnosis and management codes seem to be reasonably reliable in ascertaining people with epilepsy in routinely collected data, especially when combined with other epilepsy information such as regular prescription of anti-epileptic drugs (AEDs) or electroencephalographic (EEG) results (where available). For example, Australian, Italian, American, and UK studies achieved sensitivities of 82–94% and specificities of 94–100% when identifying people with epilepsy from routinely collected data [5–8]. The vast majority of deaths will be captured by legal frameworks involving death certificates or equivalent systems which are generally accessible for most routinely collected research systems.

    View all citing articles on Scopus
    View full text