Va

LAa

b

Fc

Ud

ARRA

KDSEISVI

0

Preventive Veterinary Medicine 114 (2014) 145–150

Contents lists available at ScienceDirect

Preventive Veterinary Medicine

j ourna l h om epa ge: www.elsev ier .com/ locate /prevetmed

alidation of the diagnosis canine epilepsy in a Swedishnimal insurance database against practice records

inda Heskea,∗, Mette Berendtb, Karin Hultin Jäderlunda, Agneta Egenvall c,ne Nødtvedtd

Department of Companion Animal Clinical Sciences, Norwegian School of Veterinary Science, Oslo, NorwayDepartment of Veterinary Clinical and Animal Sciences, Faculty of Health and Medical Sciences, University of Copenhagen,

rederiksberg, DenmarkDepartment of Clinical Sciences, Faculty of Veterinary Medicine and Animal Husbandry, Swedish University of Agricultural Sciences,ppsala, SwedenDepartment of Production Animal Clinical Sciences, Norwegian School of Veterinary Science, Oslo, Norway

a r t i c l e i n f o

rticle history:eceived 29 October 2013eceived in revised form 21 February 2014ccepted 2 March 2014

eywords:ogseizurepilepsydiopathic epilepsyymptomatic epilepsyalidation

nsurance database

a b s t r a c t

Canine epilepsy is one of the most common neurological conditions in dogs but the actualincidence of the disease remains unknown. A Swedish animal insurance database has pre-viously been shown useful for the study of disease occurrence in companion animals. Thedogs insured by this company represent a unique population for epidemiological studies,because they are representative of the general dog population in Sweden and are followedthroughout their life allowing studies of disease incidence to be performed. The databasecovers 50% of all insured dogs (in the year 2012) which represents 40% of the nationaldog population. Most commonly, dogs are covered by both veterinary care insurance andlife insurance. Previous studies have shown that the general data quality is good, but thevalidity of a specific diagnosis should be examined carefully before using the database forincidence calculations. The aim of the current study was therefore to validate the informa-tion contained in the insurance database regarding canine epilepsy. The validation focusedon the positive predictive value and the data-transfer from the veterinary practice recordsto the insurance database. The positive predictive value was defined as the proportion ofrecorded cases that actually had the disease in question. The quality of the data-transfer wasassessed by comparing the diagnostic codes in practice records to the codes in the insur-ance database. The positive predictive value of the diagnostic codes for canine epilepsy(combining “epileptic convulsions” and “idiopathic epilepsy”) in the insurance databasewas validated in a cross-sectional study where insurance claims for canine epilepsy were

compared to diagnostic information in practice records. A random sample of dogs witha reimbursed insurance claim during 2006 was included in the study sample (n = 235).Practice records were requested by mail from attending veterinarians. Two independentexaminers scrutinized all the records. All 235 dogs were coded correctly in the database asthey really had suffered seizures with or without convulsions, and the quality of the data- transfer was therefore excellent. In total, 167 dogs (71%) were classified as cases of canine

∗ Corresponding author. Tel.: +47 22964923; fax: +47 22964962.E-mail addresses: linda.heske@nmbu.no, energicaslinda@hotmail.com (L. Heske).

http://dx.doi.org/10.1016/j.prevetmed.2014.03.003167-5877/© 2014 Elsevier B.V. All rights reserved.

146 L. Heske et al. / Preventive Veterinary Medicine 114 (2014) 145–150

epilepsy according to pre-defined criteria, and the positive predictive value was thereforeconsidered relatively high. Based on these results, it was concluded that the data regardingcanine epilepsy in the insurance database can be used for further population studies.

records. We also sought to investigate the quality of the

1. Introduction

Epilepsy is one of the most common neurological condi-tions in dogs (Knowles, 1998). The disease is characterizedby recurrent seizures originating from the brain, andcaused by an imbalance between excitatory and inhibitorymechanisms (Thomas, 2010).

Definitions of human epilepsy refer to the Interna-tional League Against Epilepsy (ILAE) classification groupguidelines which are continuously under revision. Similarterminology is traditionally used in veterinary medicine,although no consensus exists (Berendt & Gram, 1999;Engel, 2006; Chandler, 2006; Berg et al., 2010). Epilepsyis classified by etiology. The most widely used veterinarydefinition of idiopathic epilepsy (IE) is epilepsy where nointracranial morphological abnormalities can be identifiedand a genetic disposition for seizures may be suspected,whereas symptomatic epilepsy (SE) is defined as caused byan identifiable structural lesion in the brain. These defini-tions have been applied in this study.

Regardless of etiology, the duration of an epilepticseizure seldom exceeds 1–2 min. Continued seizure activ-ity lasting for a longer period, without the dog regainingconsciousness, is defined as status epilepticus. Two ormore isolated epileptic seizures within less than 24 hare regarded as one seizure event/episode, often calleda cluster. Seizures with or without convulsions mayalso be provoked by extracranial causes, e.g. toxic ormetabolic disorders, but do not fulfill the definition ofepilepsy.

Epidemiologic data regarding canine epilepsy fromlarger populations are sparse and there is a need for pop-ulation based studies of the incidence. The prevalence ofcanine epilepsy has been estimated to 1–2% in a hospitalpopulation (Schwartz-Porsche, 1994). However, in breedsdisplaying genetic IE the prevalence is often much higher(Berendt et al., 2002, 2008; Gulløv et al., 2011; Casal et al.,2006).

Insurance claims files are examples of secondarydatabases, implying that the data were not originally col-lected for research purposes. A current estimate showsthat Agria (www.agria.se) covers approximately 50% of theinsured dogs, which represents 40% of the Swedish dogpopulation (M. Berglin, personal communication, 2012)and it has been shown that the population insured byAgria is comparable to the general population of dogsin Sweden (Egenvall et al., 1999). A diagnostic cod-ing system, with approximately 8.000 codes, is usedby Agria and applied uniformly by the insurance com-pany and submitting veterinary practitioners (SwedishAnimal Hospital Association, 1993). By accessing this pet-

insurance database, it has been possible to perform large-scale retrospective cohort studies among Swedish dogsto determine general as well as specific disease-incidence

© 2014 Elsevier B.V. All rights reserved.

(Egenvall et al., 2000; Bergström et al., 2006; Fall et al.,2007; Nødtvedt et al., 2006a; Bergknut et al., 2012).

Dogs insured by Agria are commonly covered by bothveterinary care insurance and life insurance. For veterinarycare insurance the cost of veterinary treatment exceedingthe deductible will be reimbursed in case of disease. A lifeinsurance plan means that the owner will be reimbursedthe monetary value of their pet if the animal dies or is euth-anized due to disease or an accident before 10 years of age.Because animals are followed throughout life, calculationof disease incidence is possible. However, it is important toestablish the diagnostic validity regarding a specific diseaseprior to using the insurance database for research.

Validation can be performed in various ways. Informa-tion in an administrative database, e.g. summary healthrecords or insurance data, will often be compared to infor-mation in medical records or another database whichtheoretically captures the same cases (Christensen et al.,2007). According to Pfeiffer (2010) “accuracy can be definedas the closeness between ‘test result’ and the ‘true clini-cal’ state”. A review by Hogan and Wagner (1997) assessedaccuracy in terms of completeness and correctness of therecords. Completeness has been defined as the propor-tion of disease cases in the population that were actuallyrecorded in the database and correctness as the proportionof recorded cases that actually had the disease in question(Jordan et al., 2004). This definition of correctness is equiva-lent to the positive predictive value (PPV): the proportion ofdogs recorded in the insurance database as having epilepsywhich actually have the disease. Furthermore, Jordan et al.(2004) lets external validity refer to whether the patientactually has the disease indicated by the diagnostic codewhereas internal validity refers to whether the patient iscoded in the database with the diagnosis believed to becorrect by the clinician. In this context, internal validitywill reflect how well the data-transfer functions betweenpractice records and the central database, while externalvalidity reflects the skills of the clinician. The accuracyof diagnostic information in the Swedish Agria insurancedatabase has previously been validated against randomlyselected practice records and considered to be reasonable(Egenvall et al., 1998), but it should be noted that only qual-ity of the data-transfer was assessed. A validation regardingcanine atopic dermatitis in the database has also been per-formed, assessing both the skills of the clinicians and thedata-transfer process (Nødtvedt et al., 2006b).

The primary objective of the present study was to vali-date the data quality regarding the diagnosis epilepsy inAgria’s insurance database by determining the positivepredictive value of the diagnosis compared to practice

data-transfer from practice records to the database. As asub-goal, it was attempted to determine if the dogs diag-nosed with epilepsy could be classified with IE or SE.

rinary M

2

2

scwprdT((eatac2

tc1tsbvnid

2

fsocldAamr

il(ficsfimtAwriw

L. Heske et al. / Preventive Vete

. Materials and methods

.1. Data collection

Life insurance and veterinary care insurance claimsettled during 2006 with either of the two diagnosticodes (“epileptic convulsions” or “idiopathic epilepsy”)ere selected from the insurance database in 2011. A sam-le of these claims was drawn using computer generatedandom numbers. It was postulated that the positive pre-ictive value would be at least 85% (Egenvall et al., 1998).he required sample size was estimated to 196 recordswith 0.05 value for alpha and a power (1-beta) of 0.95)www.openepi.com, infinite population and no designffect). The number was increased somewhat because ofnticipated non-response, and ultimately paper copies ofhe claims files from 235 dogs from the insurance databaserchives were obtained. In cases where more than onelaim was filed, only the first claim for each dog during006 was included.

Complete practice records were requested by mail fromhe attending veterinarians (n = 92) of all 235 selectedases. The number of cases from each clinic ranged from

to 21 claims, with a median of one case (55 clinics con-ributed one patient each). The veterinarians were asked toubmit all practice records for the selected cases identifiedy pet name, breed, date of birth, record number, date ofisit, diagnosis, and owner name. The veterinarians wereot informed about the specific aim of the study but were

nformed that dogs with reimbursed claims for neurologiciseases were being evaluated.

.2. Data handling

The following information was extracted and recordedrom the practice records when available: date of birth,ex, breed, age at first seizure, seizure semiology (thebservable manifestations of the seizures), and results oflinical investigations (e.g. clinical examination, neuro-ogical examination, hematology and serum biochemistry,iagnostic imaging and cerebrospinal fluid examination).dditional information included antiepileptic treatmentnd treatment monitoring. For dogs that had died, infor-ation regarding time from index seizure to death was

ecorded.Two examiners (LH and MB) evaluated all records

ndependently and classified them according to the fol-owing criteria: epilepsy was defined as recurrent seizurestwo or more) of intracranial cause. In order to qualifyor a diagnosis of IE, the dogs should have been clin-cally and neurologically normal at both the veterinaryonsultation and according to the owner between theeizures, with complete blood count and biochemistry pro-le within normal limits. Furthermore, a minimum of 4onths follow-up after the first seizure, with no indica-

ion of progressive intracranial disease, should be available. diagnosis of SE required that intracranial pathology

as confirmed by computer tomography (CT), magnetic

esonance imaging (MRI), cerebrospinal fluid (CSF) exam-nation or post-mortem histopathology. Status epilepticus

as defined as a single epileptic seizure event lasting for

edicine 114 (2014) 145–150 147

more than 30 min, or a series of seizures during whichnormal consciousness was not regained between epilepticevents over a period longer than 30 min. Cluster seizureswere defined as >1 epileptic seizure within 24 h.

Based on the criteria described above, a set of diagnosticcategories was defined. The inclusion and exclusion criteriafor each category are summarized in the flowchart pre-sented in Fig. 1 and in Table 1. Results from the two raterswere compared and any disagreements in classification ofcases were solved by consensus. Cases with a lack of dis-tinguishing information from the records were grouped asseizures of unclassifiable origin due to sparse record infor-mation. All cases were thereby assigned into ‘unambiguous’categories as shown in Fig. 1. In this paper, sub-groupswere combined to form four main groups of seizures: (1)repeated seizures of intracranial cause = canine epilepsy(including IE, SE, and non-classifiable epilepsy type), (2)single seizure events, (3) unclassifiable due to sparse recordinformation, and (4) seizures of extracranial cause. Thediagnostic work-up performed for dogs belonging to group(1) was summarized.

3. Results

The insurance database included 202 life-insuranceclaims and 1189 veterinary care claims for “idiopathicepilepsy” and “epileptic convulsions” in 2006. This rep-resents a total of 1391 individuals that had at least onereimbursed claim for epilepsy. A total of 189 completepractice records were supplied by the veterinary practi-tioners out of the 235 requested, giving a total effectiveresponse rate of 80.4%. Practice records (in some casesincomplete) that were already available in the Agriaarchives were included in order to obtain information fromall 235 cases. Of these cases, 175 were veterinary careclaims and 60 were life-insurance claims. Eighty differentbreeds were represented in this subset. The most commonbreeds were mongrel (n = 26), Labrador retriever (n = 26),Cavalier King Charles spaniel (n = 11), Golden retriever(n = 10) and German shepherd (n = 10).

All 235 dogs coded with “epileptic convulsions” or “idio-pathic epilepsy” in the database had suffered from seizureepisodes (with or without convulsions) and were alsocoded as such in the practice records. The quality of data-transfer was therefore deemed to be excellent.

Of the 235 dogs that had seizures with or without con-vulsions, 167 (71%) were evaluated as being dogs withepilepsy based on the criteria in Table 1/Fig. 1. The posi-tive predictive value (the proportion of recorded cases thatactually had the disease in question) of the diagnostic codesfor “epileptic convulsions” and “idiopathic epilepsy” in theAgria database was therefore 71% regarding epilepsy. Thedistribution of dogs after scrutiny of the practice records ispresented by diagnostic category in Fig. 1.

Of the 235 dogs with seizures with or without con-vulsions, 27 dogs (11.5%) had only one recorded seizureepisode in the practice record and thus did not meet

the criteria for epilepsy (requiring two or more seizures)applied in the study (Table 1/Fig. 1). In 39 dogs (16.6%) thecause of seizures were not sufficiently documented in thepractice records and thus did not qualify as epilepsy cases.

148 L. Heske et al. / Preventive Veterinary Medicine 114 (2014) 145–150

Fig. 1. Flowchart illustrating classification of 235 Swedish dogs with a reimbursed insurance claim for “epileptic convulsions” or “idiopathic epilepsy”during 2006. Non-classifiable epilepsy type includes dogs with epilepsy but not enough information to sub-classify into idiopathic epilepsy or symptomaticepilepsy.

Table 1Explanation of the categories from flowchart in Fig. 1.

Categories in the flowchart Explanation

Seizures with or without convulsions Dogs with diagnostic codes “epileptic convulsions” or “idiopathic epilepsy”that had suffered seizures.

Single event of status epilepticus, clusterseizures within 24 h or one single seizure

Dogs with one single episode of status epilepticus, cluster seizures within 24 hor only one single seizure with different time of follow-up according to theavailable veterinary record.

Single event ≥4 months follow-up Single event with at least 4 months follow-up.Single event <4 months follow-up Single event with less than 4 months follow-up.Single event, spontaneous death Single event where the patient died during or in connection with the seizure.Single event, euthanasia Single event where the patient was euthanized as a consequence of the seizure

or because of poor prognosis, i.e. severe status epilepticus.

Seizures that are unclassifiable due to sparserecord info

Patients with seizures with or without convulsions in the record but with nofurther information to put the patient in one of the other categories.

Repeated seizures of intracranial cause; canine epilepsyIdiopathic epilepsy The IE patients had an age of onset ≥6 month ≤7 years old and a follow up

time of ≥4 months after the first seizure with no other neurological signsdeveloping. Information of IE in close relatives was also considered importantfor the classification into the IE category.

Non-classifiable epilepsy type Cases which fulfilled the criteria for canine epilepsy but for some reason wasunable to sub-classify to either IE or SE.

Symptomatic epilepsy For SE patients, persistent neurological signs reported in the records,morphological changes on brain diagnostic imaging and an age of onset <6months or ≥7 years were the most important factors for the classification.

Repeated seizures of extracranial cause Patients with seizures of extracranial cause.

L. Heske et al. / Preventive Veterinary M

Table 2The clinical work-up done by attending veterinarians and reported in therecords for 167 dogs with reimbursed claims for “epileptic convulsions”or “idiopathic epilepsy” during 2006 with practice records which allowedclassification as epileptic.

Clinical work up No. of dogs Percentage (%)

Clinical examination 131 78.4Neurological

examination14 8.4

Routine blood analysis 126 75.4Urinalysis 13 7.8Diagnostic imaging of

the brain CTa/MRb17 10.2

AEMc 115 68.9Monitoring of AEM 70 41.9Necropsy 4 2.4

a

Ite

((fiSH(r

4

eIeb“fifTrttew7traDpf

eo2eri

Computer tomography.b Magnetic resonance imaging.c Antiepileptic medication.

n two dogs (0.9%) the practice records documented thathe cause of the seizures were extracranial (and thus notpilepsy).

Of the 167 dogs evaluated as “true” epilepsy cases, 8651.5%) were classified as IE and 5 (3%) as SE. For 76 cases45.5%) the information in the veterinary records was insuf-cient to attempt classification of epilepsy type into IE orE. The classification of epilepsy type is shown from Fig. 1.ow the attending veterinarians diagnosed the 167/234

71%) patients with epilepsy, according to the practiceecords, is shown in Table 2.

. Discussion

A validation of the data quality regarding caninepilepsy in a Swedish insurance database was performed.n terms of the usefulness of the database for studies ofpilepsy in dogs, the data-transfer was excellent (100%)etween the diagnostic codes (“epileptic convulsions” andidiopathic epilepsy”) retrieved from the insurance claimsles and the codes assigned by the clinicians and retrieved

rom the practice records from the sampled population.his reflects the fact that data-transfer between practiceecords and the central database does not introduce erroro any large extent. However, when estimating the posi-ive predictive value versus specific diagnostic criteria forpilepsy the picture changes. The positive predictive valueas considered relatively high as we could conclude that

1% of the cases did indeed have epilepsy. This reflectshe skills of the clinicians as documented in the practiceecords which were scrutinized. Our result is supported by

validation study of epilepsy diagnoses in humans using aanish national hospital register which found that 81% ofersons registered with epilepsy fulfilled the ILAE criteriaor a diagnosis of epilepsy (Christensen et al., 2007).

In the present study we chose not to define dogs aspileptic if they had only one known single seizure orne episode of status epilepticus or cluster seizures within

4 h, as these cases were considered as a single epilepticvent. In some of these cases it is reasonable to expect thatepeated seizure episodes could happen in the future, andf they were followed in a longitudinal study this could

edicine 114 (2014) 145–150 149

potentially increase the number of verified cases ofepilepsy. However, it was beyond the scope of the currentstudy to follow dogs through time or to gather additionalinformation from clinicians beyond the original practicerecords. Given the strict implementation of this inclusioncriterion, and also because we expect some epileptic dogsamong those having seizures unclassifiable due to sparserecord information, it seems reasonable to assume that thefrequency of 71% of epilepsy in this population is underes-timated.

An optimally performed validation study should alsoinclude validation of sensitivity (the proportion of diseasecases in the population that were actually recorded) of theepilepsy diagnosis in the database, but due to the natureand the retrospective form of the study this was not feasi-ble to achieve. Estimating sensitivity would require accessto some other database containing information about the“true” number of epileptic dogs in the Swedish dog popu-lation. Digitalized practice record systems from veterinaryclinics’ could have been used for this purpose, but wouldstill not provide a complete list of cases. Furthermore, itis possible that dogs without an epilepsy diagnosis in theAgria insurance database could have had epileptic seizuresthat were not recorded, mainly due to not reaching thedeductible.

A previous study of general claims in small animalpractice (Egenvall et al., 1998) found that the diagnosiswas identical in the insurance database and the practicerecords in of 85% of the claims. Nødtvedt et al. (2006b) val-idated the diagnosis canine atopic dermatitis and found theproportion identical diagnosis in the Agria database to be100% indicating that data-transfer was working optimally.However, the clinicians did not always follow the publishedguidelines for making the diagnosis, resulting in between84.2% (liberal interpretation of records) and 40.9% (strictinterpretation of records) of dogs being classified as havingcanine atopic dermatitis after scrutiny of practice records.It should be noted that Egenvall et al. (1998) used only themedical record for the day of the claim for the validation,while Nødtvedt et al. (2006b) included the entire medicalhistory of each dog. Also, many clinics went from manualto electronic submission of claims between 1998 and 2006which probably resulted in more reliable data-transfer.

Because most epileptic seizures last less than a cou-ple of minutes, veterinarians seldom observe the actualseizure, unless the patient is hospitalized during ongo-ing status epilepticus or cluster seizures. Consequently,the diagnosis of canine epilepsy is heavily dependent onhistoric information collected from the owner, includinga careful description of seizure semiology, and backed upwith supplemental diagnostic tests. Our results regardingclinical investigations used by the attending veterinariansto assess the patients in this study reflect this. Accord-ing to the practice records from the veterinarians theirclinical investigations varied from only history takingand, for the vast majority of the cases, a clinical exam-ination and routine blood analysis. Some patients also

underwent a neurological examination and advanced diag-nostic imaging to identify a potential intracranial lesion orextracranial cause of seizures. The number of clinical testsincluded in the investigation reflects the veterinarian’s

erinary M

150 L. Heske et al. / Preventive Vet

experience, the owner’s beliefs and economy and availablediagnostic equipment. Even in human medicine, wherediagnostic methods often are more refined and cost limita-tions rarely hinder the diagnostic process, the diagnosis ofepilepsy may be questionable in some cases (Christensenet al., 2007). However, despite this limitation, the resultsof the current study showed that a minimum of 71% ofthe claims for epilepsy were correct, which is acceptable,and it can be concluded that the Agria database can beused for further epidemiologic studies of canine epilepsy.It should, however, be kept in mind that the incidence ofepilepsy may be under-estimated in the insured populationin general because the cost of veterinary care for epilepsy insome instances may not exceed the deductible in additionto the inherent (conservative) misclassification describedhere.

A classification of epileptic dogs into subgroups of IEor SE was possible for 91/167 (54.5%) of the cases. Thisleaves behind a group with non-classifiable epilepsy typeconstituting 76/167 (45.5%) of the epilepsy cases. Not sur-prisingly, distinguishing IE from SE by veterinary recordswritten in a clinical setting at a regular veterinary practiceproved to be challenging.

Of the epileptic dogs (n = 167), few were placed in theSE group (3%). Evidence of an extracranial cause of theseizures was apparent in less than 1% of all the includeddogs (n = 235). A possible explanation may be that only twodiagnostic codes were selected for this study; “epilepticconvulsions” and “idiopathic epilepsy”. Because there aremore specific codes available for the underlying diagnosesof extra cranial causes of seizures or SE, e.g. insulinoma orbrain neoplasia: these codes were probably used in manysuch cases excluding them from the current study. On theother hand, the 16.6% of all included dogs (n = 235) withseizures unclassifiable due to sparse record information,as well as the 45.5% of epileptic dogs (n = 167) grouped asnon-classifiable epilepsy type, could also hide a larger pro-portion of any sub-category of seizuring dogs and hence thepresented proportions in the sub-categories are somewhatuncertain.

5. Conclusion

The diagnostic codes “epileptic convulsions” and “idio-pathic epilepsy” in the Agria database were validatedagainst information found in practice records, and thequality of data-transfer process was deemed excellent.When applying strict criteria for canine epilepsy, the pos-itive predictive value of the diagnosis epilepsy in thedatabase was at least 71%. Based on these results, it wasconcluded that the data regarding canine epilepsy in theinsurance database can be used for further population stud-ies regarding this condition.

Acknowledgements

The authors wish to thank Agria Insurance for supplyingdata and the Agria Insurance Foundation for Research forfinancial support. Thank you also to all the veterinary clin-ics throughout Sweden that responded to our request and

edicine 114 (2014) 145–150

made their practice records available. This help is greatlyappreciated.

References

Berendt, M., Gram, L., 1999. Epilepsy and seizure classification in 63 dogs:a reappraisal of veterinary epilepsy terminology. J. Vet. Intern. Med.13, 14–20.

Berendt, M., Gredal, H., Pedersen, L.G., Alban, L., Alving, J., 2002.A cross-sectional study of epilepsy in Danish Labrador retriev-ers: prevalence and selected risk factors. J. Vet. Intern. Med. 16,262–268.

Berendt, M., Gulløv, C.H., Christensen, S.L.K., Gudmundsdottir, H., Gredal,H., Fredholm, M., Alban, L., 2008. Prevalence and characteristics ofepilepsy in the Belgian shepherd variants Groenendael and Tervuerenborn in Denmark 1995–2004. Acta Vet. Scand. 50, 51.

Berg, A.T., Berkovic, S.F., Brodie, M.J., Buchhalter, J., Cross, H.J., van EmdeBoas, W., Engel, J., French, J., Glauser, T.A., Mathern, G.W., Moshe, S.L.,Nordli, D., Plouin, P., Scheffer, I.E., 2010. Revised terminology and con-cepts for organization of seizures and epilepsies: report of the ILAECommission on Classification and Terminology, 2005–2009. Epilepsia51, 676–685.

Bergknut, N., Egenvall, A., Hagman, R., Gustås, P., Hazewinkel, H.A.,Meij, B.P., Lagerstedt, A.S., 2012. Incidence of intervertebral diskdegeneration-related diseases and associated mortality rates in dogs.J. Am. Vet. Med. Assoc. 240, 1300–1309.

Bergström, A., Nødtvedt, A., Lagerstedt, A.S., Egenvall, A., 2006. Incidenceand breed predilection for dystocia and risk factors for cesareansection in a Swedish population of insured dogs. Vet. Surg. 35,786–791.

Casal, M.L., Munuve, R.M., Janis, M.A., Werner, P., Henthorn, P.S., 2006.Epilepsy in Irish Wolfhounds. J. Vet. Intern. Med. 20, 131–135.

Chandler, K., 2006. Canine epilepsy: what can we learn from humanseizure disorders? Vet. J. 172, 207–217.

Christensen, J., Vestergaard, M., Olsen, J., Sidenius, P., 2007. Validation ofepilepsy diagnoses in the Danish National Hospital Register. EpilepsyRes. 75, 162–170.

Egenvall, A., Bonnet, B.N., Olson, P., Hedhammar, Å., 1998. Validation ofcomputerized Swedish dog and cat insurance data against veterinarypractice records. Prev. Vet. Med. 36, 51–65.

Egenvall, A., Hedhammar, Å., Bonnett, B.N., Olson, P., 1999. Survey of theSwedish dog population: age, gender, breed, location and enrolmentin animal insurance. Acta Vet. Scand. 40, 231–240.

Egenvall, A., Bonnett, B.N., Olson, P., Hedhammar, A., 2000. Gender, age,breed and distribution of morbidity and mortality in insured dogs inSweden during 1995 and 1996. Vet. Rec. 146, 519–525.

Engel, J., 2006. ILAE classification of epilepsy syndromes. Epilepsy Res. 70(Suppl), 5–10.

Fall, T., Hamlin, H.H., Hedhammar, Å., Kämpe, O., Egenvall, A., 2007.Diabetes mellitus in a population of 180,000 insured dogs: inci-dence, survival, and breed distribution. J. Vet. Intern. Med. 21,1209–1216.

Gulløv, C.H., Toft, N., Baadsager, M.M., Berendt, M., 2011. Epilepsy in thePetit Basset Griffon Vendeen: prevalence, semiology, and clinical phe-notype. J. Vet. Intern. Med. 25, 1372–1378.

Hogan, W.R., WagnerF M, 1997. Accuracy of data in computer-basedpatient records. J. Am. Med. Inform. Assoc. 4, 342–355.

Jordan, K., Porcheret, M., Croft, P., 2004. Quality of morbidity coding ingeneral practice computerized medical records: a systematic review.Fam. Pract. 21, 396–412.

Knowles, K., 1998. Idiopathic epilepsy. Clin. Tech. Small Anim. Pract. 13,144–151.

Nødtvedt, A., Egenvall, A., Bergvall, K., Hedhammar, A., 2006a. Incidenceof and risk factors for atopic dermatitis in a Swedish population ofinsured dogs. Vet. Rec. 159, 241–246.

Nødtvedt, A., Bergvall, K., Emanuelson, U., Egenvall, A., 2006b. Canineatopic dermatitis: validation of recorded diagnosis against practicerecords in 335 insured Swedish dogs. Acta Vet. Scand. 48, 8.

Pfeiffer, D. (Ed.), 2010. Veterinary Epidemiology. An Introduction.Wiley–Blackwell, Oxford.

Schwartz-Porsche, D., 1994. Seizures. In: Braund, K.G. (Ed.), Clinical Syn-dromes in Veterinary Neurology. , 2nd ed. Mosby, St. Louis, MO, pp.234–251.

Swedish Animal Hospital Association, 1993. Diagnostic registry for thehorse, the dog and the cat (Diagnosregister för häst, hund och katt)Tåberg. Tåbergs tryckeri, Sweden.

Thomas, W.B., 2010. Idiopathic epilepsy in dogs and cats. Vet. Clin. NorthAm. Small Anim. 40, 161–179.