Estimating the incidence of first unprovoked seizure and newly diagnosed epilepsy in the low-income urban community of Northern Manhattan, New York City


  • During the conduct of this research Dr. Shih was with the Department of Neurology and Dr. Leary was with the Department of Pediatrics and the Department of Neurology, Columbia University College of Physicians & Surgeons, New York, New York, U.S.A.

Address correspondence to Dale C. Hesdorffer, Ph.D., GH Sergievsky Center, Columbia University, P & S Unit 16, 630 West 168th Street, New York, New York, 10032, U.S.A. E-mail:


Purpose: To estimate the incidence and mortality associated with first unprovoked seizure or newly diagnosed epilepsy in a low-income, predominantly Hispanic community in Northern Manhattan, New York City.

Methods: We performed a population-based study to determine the incidence of first unprovoked seizure or newly diagnosed epilepsy. Participants were Northern Manhattan residents seen at area hospitals and nursing homes between 2003 and 2005. Cumulative probability of mortality and standardized mortality ratios (SMRs) were also calculated.

Results: Among 209 incident cases identified, 123 (58.9%) presented with an incident single unprovoked seizure. A total of 138 (66.0%) participants were Hispanic and 94 (45.0%) had a median household income under $15,000/year. The overall age and sex-adjusted incidence of all unprovoked seizures was 41.1 (95%CI = 35.4–46.8) per 100,000 person-years. Higher incidence was observed in low-income groups. Incidence among Hispanics was similar to that of non-Hispanic whites and non-Hispanic blacks. The cumulative probability of mortality was 17% (95%CI = 12–24%) by 3 years after diagnosis and was significantly greater in females and in those with an identified etiology. SMRs were significantly increased for all groups with respect to age, Hispanic ethnicity, middle and high income, partial seizure type, and remote symptomatic etiology. Idiopathic/cryptogenic and progressive symptomatic etiologies, low income, gender, and non-Hispanic ethnicity were not associated with a significantly increased SMR.

Conclusion: Incidence of first unprovoked seizure or newly diagnosed epilepsy did not differ by race-ethnicity. Although lower income was associated with higher incidence, higher income was associated with an increased SMR. Future research should examine reasons for differential incidence by income.

There are scarce data regarding the incidence of single unprovoked seizures and newly diagnosed epilepsy in minority populations and in poor urban communities in the United States. Prevalence studies have predominated in minority populations, finding a higher prevalence of epilepsy among African-Americans than among white populations (Haerer et al., 1986; Cowan et al., 1989; Murphy et al., 1995), but results among Hispanic populations have been mixed (Cruz et al., 1985; Lavados et al., 1992; Placencia et al., 1992; Basch et al., 1997; Nicoletti et al., 1999; Medina et al., 2005).

In studies of predominantly white populations from developed countries, the age-adjusted incidence of first unprovoked seizure and newly diagnosed epilepsy is 48 per 100,000 person-years (Annegers et al., 1999) to 69 per 100,000 person-years (Hauser et al., 1993) and the incidence of epilepsy ranges from 26 per 100,000 person-years (de Graaf, 1974) to 51 per 100,000 person-years (Hauser et al., 1993). The cumulative incidence of epilepsy is 1.2% through age 24 and 3.1% through age 74 (Hauser et al., 1993; Hauser et al., 1996).

We undertook a population-based study to estimate the incidence of first unprovoked seizures and newly diagnosed epilepsy among a low-income, mostly Hispanic community in a major urban setting. Our study aimed to identify all incident-unprovoked seizures and epilepsy among the population residing in the communities of Washington Heights and Inwood, New York City. Additionally, we determined the mortality of the cohort over a median follow-up period of 2.9 years.


We created an ascertainment system at hospitals in the northern part of Manhattan and in the Bronx as well as in hospitals and nursing homes in the Northern Manhattan communities of Washington Heights and Inwood to identify incident unprovoked seizure and newly diagnosed epilepsy over a 2-year period in order to determine the incidence of epilepsy and first unprovoked seizure. In accordance with the guidelines set forth by the International League Against Epilepsy (ILAE), we defined unprovoked seizure as a seizure without an identified proximate precipitant or multiple such seizures within a 24-h period, and epilepsy as recurrent unprovoked seizures. These diagnoses exclude individuals with only febrile seizures, neonatal seizures, or acute symptomatic seizures.

Study subjects

Patients were considered to be incident if they experienced their first unprovoked seizure or were newly diagnosed with epilepsy between June 1, 2003 and June 1, 2005 and resided in the Northern Manhattan communities of Washington Heights and Inwood, defined by the zip codes 10031, 10032, 10033, 10034, and 10040. These communities form the northern end of Manhattan, New York City, and represent the catchment area for the New York Presbyterian Hospital (NYPH) and its affiliated community hospital, The Allen Pavilion. While pilot data had suggested that the overwhelming majority of people with a first single unprovoked seizure or newly diagnosed epilepsy (cases) would be identified from these hospitals, we enlarged our seizure surveillance system to include the two nursing homes in the community and four hospitals outside the community, two in upper Manhattan (St. Luke's Roosevelt Hospital and Harlem Hospital) and two in the Bronx (Lincoln Hospital and North Central Bronx Hospital). These hospitals were selected based upon data from the New York State-wide Planning and Research Cooperative System (SPARCS) that records discharge diagnosis of hospitalized patients. SPARCS showed that seizure and epilepsy patients from Northern Manhattan who were not seen at NYPH or the Allen Pavilion were most likely to be hospitalized at the four selected hospitals outside the community. There were no private neurologists practicing in the community outside of those affiliated with NYPH.

We set up a local seizure surveillance system at NYPH and The Allen Pavilion to ascertain incident cases. Surveillance included daily screening of the Pediatric and Adult Emergency Department logs for the following chief complaints at admission: seizure, convulsion, fainted, unresponsive, loss of consciousness, falls, dizziness, paresthesias, numbness, syncope, near syncope, and unilateral weakness. Additional surveillance included regular contact with the neurology consult residents, and a weekly review of lists of hospital discharges with ICD-9 codes 780.3, and 345.0–345.9. Additionally, we screened the Epilepsy and Neurology clinics, the Epilepsy Monitoring Unit and the epileptologist's private offices for incident cases. After review of medical records for potential incident cases, those judged to have experienced a first unprovoked seizure or to have been newly diagnosed with epilepsy received a letter signed by the ED director or their attending physician, explaining the study. After 2 weeks, if the potential case had not called to refuse, they were telephoned by study staff to confirm eligibility and to invite eligible individuals to participate. Cases who agreed to participate were interviewed about their seizure phenomenology, demographics, use of medical care, comorbidities, quality of life, stigma associated with epilepsy, and whether adults of working age were employed. Additionally, parents of children with new onset seizures were interviewed about parental stress. All cases were followed every 4 months for a year for further seizures, use of medical care, quality of life, stigma, parental stress and paid work. If the case indicated that they did not want to be contacted, could not be contacted after repeated attempts or had died prior to the contact, then the medical record was abstracted for the seizure information.

At the four hospitals located outside the Washington Heights and Inwood communities, we screened the ED logs in the same way as at NYPH. At the two nursing homes in the community, we screened medical records to identify incident cases, utilizing the same criteria for those cases assessed in the ED in addition to screening for use of an antiepileptic drug. Potential cases identified at the nursing homes and the four additional hospitals were not offered participation in the interview portion of the study due to funding constraints. Instead, we reviewed and abstracted medical records.

This study was approved by the IRB at each institution and the CDC. A certificate of confidentiality was issued by the NIH.

Measures and assessments


Age at the first unprovoked seizure or newly diagnosed epilepsy was categorized as <1-year old, 1–4 years old, 5–9 years old, 10–14 years old, 15–24, 25–34, 35–44, 45–54, 55–64, 65–74, 75–84, and 85 years and older.

Race and ethnicity

Interviewed participants were asked about ethnicity and race. Answers to these questions were compared to the self-designated information from the medical records. Using the interview responses as the gold standard, sensitivity and specificity of the self-designation recorded in the medical record was very good. Thus, the medical record classification was used for all participants who were not interviewed. Race-ethnicity was analyzed as Hispanic, non-Hispanic white, and non-Hispanic black. Cases falling outside of this classification were categorized as Other.

Annual household income

Because the community is predominantly of very low income, household income was categorized among interviewed cases as: less than $15,000 per year, $15–49,999 per year, and $50,000 and greater. We assigned income levels to the cases, who were not interviewed, by randomly assigning them to income categories based upon the distribution of these categories in the census population of Washington Heights and Inwood. This approach was conservative because the income from the census was higher than that of the interviewed cases.

Seizure characteristics

Study epileptologists (WAH, TS, LL) classified seizures by etiology, seizure type, and epilepsy syndrome in consensus conferences. All available information was considered in these classifications, including the medical record, interview, EEG, and brain imaging. More than two-thirds (68.7%) of the patients who had brain imaging (N = 198) and 71.4% of those who had an EEG (N = 105) were interviewed.

Seizure etiology

Etiology was classified as remote symptomatic, progressive symptomatic, idiopathic/cryptogenic, or unclassifiable if there was inadequate information (ILAE, 1993). Remote symptomatic seizures included unprovoked seizures associated with static neurological disorders such as remote central nervous system infection, traumatic brain injury, cerebrovascular disease, mental retardation, cerebral palsy and autism. Progressive symptomatic seizures included unprovoked seizures associated with progressive neurological disorders such as Alzheimer's disease, primary or secondary brain neoplasm, multiple sclerosis and Parkinson's disease. When no clear antecedent was identified, unprovoked seizures were classified as idiopathic/cryptogenic.

Seizure type

Seizure type was in accordance with guidelines from the ILAE. Partial seizures included simple partial seizures, complex partial seizures, secondary generalized seizures, and undetermined focal seizures when evidence was unclear concerning the state of consciousness. Generalized seizures included primary generalized, tonic, clonic, tonic–clonic, absence, myoclonic, and those for which there was insufficient information to determine whether or not the seizure had a focal onset. Generalized tonic–clonic seizures that occurred during sleep fell into the latter category. Seizures were classified as focal when evidence existed to determine a focal onset. A seizure was considered unclassifiable when classification was impossible due to inadequate information.


Death information for the sample and for the 2004 Washington Heights and Inwood population was provided by the New York City Department of Health and Mental Hygiene. The National Death Index was not used because recorded deaths are more than 2 years behind whereas the Department of Health records were up to date. Death information for the 2004 US population was ascertained through the National Center for Health Statistics (

Statistical analysis

For descriptive data, comparisons were made using Student's t-tests for continuous data and chi-square for categorical data. Data were analyzed using SAS 9.0 (SAS Institute, Cary, NC, U.S.A.).

Incidence: Incidence was defined as the number of cases of newly diagnosed epilepsy and first unprovoked seizure within the specified population over the 2-year study period. Average annual incidence was defined as the number of new cases divided by the population at risk, expressed as a rate per 100,000 person-years, except for incidence by income where it was expressed as a rate per 100,000 household-years. The community population was determined with data from the 2000 United States Census for the zip codes that define the population of Washington Heights and Inwood, New York City. The total person-years of observation were 541,358.

The census records household income for the communities of Washington Heights and Inwood. Household income was queried in the study cohort, because this, rather than individual income, was thought to better reflect a person's socioeconomic status. Income-specific incidence was reported per 100,000 household-years. The total household-years of observation were 184,976.

Age- and sex-adjustment was performed using the direct method with the total US population from the 2000 Census as the standard. Incidence was calculated overall and by sex, age, race-ethnicity, annual household income, seizure type, and seizure etiology. We calculated 95% confidence intervals according to the method described by Breslow and Day (Breslow & Day, 1987).

Cumulative incidence

Cumulative incidence was calculated by multiplying the sum of the age-specific incidences by the study period of 2 years. This method was also used to calculate cumulative incidence among Hispanics, non-Hispanics, non-Hispanic blacks, non-Hispanic whites, and non-Hispanic Other.


Cumulative probability of mortality was calculated according to the Kaplan–Meier method; 95% confidence intervals were also determined. The log rank statistic was used to test the significance of the stratified estimates.

Standardized mortality ratios (SMRs) were calculated, comparing the number of observed deaths to the number expected based on death rates from a standard population (Kelsey et al., 1986). SMRs were separately calculated using the 2004 Washington Heights and Inwood death rates and the 2003 US population death rates as the standard. Death rates for both standard populations were assumed to be constant for the median follow-up period of 2.9 years.


First unprovoked seizures and newly diagnosed epilepsy were identified in 209 residents of Washington Heights and Inwood during the study period. Of these cases, 123 (58.9%) presented with a single unprovoked incident seizure and 82 (39.2%) presented with newly diagnosed epilepsy; the distinction was unclear in four patients (2.0%). Most cases were male (Table 1). Similar to the Northern Manhattan community in which the cases lived, most cases were of Hispanic origin (Table 1). The median age (interquartile range) of the cases was 37.4 years (14.6–68.4 years). The median age was 24.5 years (IQR = 10.7–55.9 years) among Hispanic cases and 54.6 years (IQR = 21.6–81.0 years) among non-Hispanics (Wilcoxon rank sum p < 0.001). Corresponding figures for the Washington Heights and Inwood community were 32.4 years (17.1–49.1 years) overall, 30.3 (15.3–46.4 years) among Hispanics, and 37.5 years (22.5–55.0 years) among non-Hispanics.

Table 1.  Incidence of unprovoked seizures by population characteristics
 N (%)Population (%)Incidence per 100,000 (95% CI)Age and sex-adjusted incidence per 100,000 (95% CI)
  1. aAge-adjusted only.

  2. bPopulation is total number of households in Northern Manhattan with an annual household income of under $15,000, between $15,000 and $49,999, and $50,000 or more. The incidence rate is reported per 100,000 household-years.

  3. cFour cases were unclassifiable (2.0%).

  4. dThree cases were unclassifiable (1.4%).

  5. eEight cases were unclassifiable (3.8%).

 Total209270,677 38.6 (33.4–43.8)41.1 (35.4–46.8)
 Male113 (54.1%) 129,038 (47.7%)43.8 (35.7–51.9) 46.6 (37.8–55.4)a
 Female96 (45.9%)141,639 (52.3%)33.9 (27.1–40.7) 35.9 (28.7–43.1)a
 Hispanic138 (66.0%)191,033 (70.6%)36.1 (30.1–42.1)36.5 (30.2–42.8)
 Non-Hispanic white 24 (11.5%)  30,073 (11.1%)39.9 (23.9–55.9)39.4 (22.5–56.3)
 Non-Hispanic black 35 (16.8%)  39,037 (14.4%)44.8 (30.0–59.7)37.6 (23.7–51.5)
 Non-Hispanic other12 (5.7%)10,534 (3.9%)57.0 (24.7–89.2)50.3 (21.4–71.7)
 Under $15,000 94 (45.0%)28,633 (30.9%) 164.1 (131.0–197.3)
 $15,000–49,999 87 (41.6%)41,179 (44.5%)105.6 (83.4–127.8)
 $50,000 & over 28 (13.4%)22,676 (24.6%)61.7 (38.9–84.6)
Number of seizuresc
 Single seizure123 (58.9%)270,677 (100%)22.7 (18.7–26.7)23.9 (19.6–28.2)
 Epilepsy 82 (39.2%)270,677 (100%)15.1 (11.9–18.4)16.4 (12.8–20.0)
 Idiopathic/cryptogenic113 (54.1%)270,677 (100%)20.9 (17.0–24.7)21.2 (17.2–25.2)
 Remote symptomatic 80 (38.3%)270,677 (100%)14.8 (11.5–18.0)16.6 (12.9–20.3)
 Progressive symptomatic13 (6.2%)270,677 (100%)2.4 (1.1–3.7) 2.8 (1.3–4.3) 
Seizure typee
 Generalized 87 (41.6%)270,677 (100%)16.1 (12.7–19.4)16.6 (13.1–20.1)
 Partial114 (54.6%)270,677 (100%)21.1 (17.2–24.9)23.0 (18.7–27.3)

Most cases (N = 184, 88.0%) were identified at NYPH and its community hospital, The Allen Pavilion. Hispanics accounted for 68% of cases at NYPH and 52% of cases identified at the other hospitals and nursing homes (p = 0.1). There was no difference in mean age for cases identified at NYPH (40.1 ± 29.8) versus other hospitals and nursing homes (43.8 ± 29.1; p = 0.6).

Among the 133 cases identified at NYPH who were interviewed and had information on household income, 53% (N = 70) were living in households with an annual income less than $15,000 and 40% (N = 53) were living in households with an annual income between $15,000—and $49,000. Greater than half (57%) of Hispanics (N = 56) and 41% of non-Hispanics (N = 14) had a household income less than $15,000 (p = 0.12). An annual household income of $15,000–49,000 was reported in approximately 40% of Hispanics (N = 40) and 38% non-Hispanics (N = 13; p = 0.82). Three percent of Hispanics (N = 3) and 21% of non-Hispanics (N = 7) had an income of $50,000 or more (p = 0.003 by Fisher's exact test). Census data for the community indicated that 30.9% of households (N = 28,633) in Northern Manhattan had an annual household income less than $15,000, 44.5% (N = 41,179) between $15,000 and $49,000, and 24.6% (N = 22,676) had an annual household income of $50,000 or more. When cases who were not interviewed were randomly assigned to income categories based upon data from the census regarding the distribution of annual household income in Northern Manhattan, the distribution of income among all cases was 45.0% with annual incomes under $15,000, 41.6% between $15,000 and $49,000, and 13.4% with $50,000 or more.

Among the 106 Hispanic cases identified at NYPH who were interviewed, 77.4% were Dominican (N = 82), 4.7% (N = 5) were Cuban, 1.9% (N = 2) were Mexican, and 10.4% (N = 11) were Puerto Rican. 5.7% (N = 6) had unknown ethnic origin.

Seizure etiology was idiopathic/cryptogenic among 54.1% of patients (N = 113), remote symptomatic among 38.3% (N = 80), progressive symptomatic among 6.2% (N = 13), and unknown among 1.4% (N = 3). Seizures were generalized in 87 cases (41.6%) and partial in 114 cases (54.6%), seizure type could not be determined in 8 cases (3.8%). Generalized tonic–clonic seizures in which the presence or absence of focality could not be determined accounted for 97.7% (N = 85) of all generalized seizures; the remaining two were primary generalized.

Overall incidence

The incidence of first unprovoked seizure and newly diagnosed epilepsy for Washington Heights and Inwood is presented in Table 1. The crude incidence of unprovoked seizures and epilepsy was 38.6 cases per 100,000 person-years (95% CI, 33.4–43.8); the age and sex adjusted incidence was 41.1 (35.4–46.8). The age- and sex-adjusted incidence was 23.9 (19.6–28.2) for single unprovoked seizure and 16.4 (12.8–20.0) for epilepsy.

Age-specific incidence

The incidence of unprovoked seizure and epilepsy by age category for the total population is presented in Fig. 1. Incidence was highest in the first year of life and among those 75 years and older. This pattern was seen for males and females. Incidence for children less than 1-year-old was 134.4/100,000 (51.1–217.7), decreasing to 50.4/100,000 (24.9–75.9) for children 1–4 years old, and continuing to decrease across age categories until the fourth decade of life. The highest incidence rate among females occurred in the 85 years and older age group (235.5; 95% CI 123.6–347.5), although in the male population the highest incidence rate was found among the 75–84-year olds. Incidence was lowest among 25–34-year olds, and this was true for both sexes (Table 2).

Figure 1.

Age-specific incidence of all unprovoked seizures in Washington Heights and Inwood, New York City 2001–2003. In four cases, it was not possible to determine if they had a single unprovoked seizure or epilepsy. Therefore, they were excluded from the curves for these groups, but were included in the total unprovoked seizure curve.

Table 2.  Age-specific incidence (95% CI) per 100,000 person-years by gender for all unprovoked seizures
Age (years)FemalesMalesTotal
Under 1 165.7 (33.1–298.3)  104.7 (2.1–207.3)  134.4 (51.1–217.7)  
1–434.5 (4.3–64.8)   65.4 (24.9–106.0) 50.4 (24.9–75.9)  
5–929.9 (6.0–53.8)  52.7 (21.6–83.9) 41.5 (21.8–61.3)  
10–1415.5 (–2.0–32.9)   45.8 (15.9–75.6) 30.7 (13.3–48.1)  
15–2428.9 (12.6–45.3) 50.0 (29.1–70.9) 39.8 (26.4–53.1)  
25–3410.8 (1.3–20.2)  15.5 (4.0–27.0)  13.1 (5.7–20.5)   
35–4416 (4.2–27.9)  26.8 (11.0–42.7) 21.3 (11.4–31.1)  
45–5425 (8.7–41.3)  39.0 (16.9–61.1) 31.5 (18.0–44.9)  
55–6429.5 (7.6–51.3)  42.7 (13.1–72.3) 35.3 (17.4–53.2)  
65–7446.2 (14.2–78.3) 43.8 (5.4–82.2)  45.3 (20.7–69.9)  
75–84111.8 (51.0–172.6) 212.6 (92.3–332.9) 144.7 (88.0–201.5)  
85 and over279.2 (137.9–420.5)108.3 (–41.8–258.5)235.5 (123.6–347.5) 

Gender-specific incidence

The age-adjusted incidence among males (46.6, 95% CI = 37.8–55.4) was higher than that among females (35.9, 95% CI = 28.7–43.1). The overall shape of the incidence curve by age was similar for both males and females, except in the oldest age category. Compared to males, incidence was consistently lower for females in all age groups except in the lowest age group and the 85 and over age group where incidence was highest for females. Among males, incidence peaked in the 74–85-year age category (Table 2).

Validity of race-ethnicity assignment

The validity of the medical records in identifying cases' ethnicity (Hispanic vs. non-Hispanic) was evaluated by comparing ethnicity in the medical records with that attained from interviews, the gold standard. The sensitivity of the medical records was 90.6% for Hispanics and 100% for non-Hispanics. The specificity was 100% for Hispanics and 90.6% for non-Hispanics. The positive predictive value of the medical records was 100% for Hispanics and 75.6% for non-Hispanics.

Race-ethnicity specific incidence

The age- and sex-adjusted incidence rate was 36.5 (30.2–42.8) for Hispanics, 39.4 (22.5–56.3) for non-Hispanic whites, and 37.6 (23.7–51.5) among non-Hispanic blacks (Fig. 2).

Figure 2.

Incidence of all unprovoked seizures by age for Hispanics and Non-Hispanics.

Incidence by household income

Using household-years as the denominator, we observed an incidence of 164.1 (131.0–197.3) per 100,000 household-years among the lowest income group, 105.6 (83.4–127.8) among the middle-income group, and 61.7 (38.9–84.6) among those with an annual income of $50,000 or more (Table 1).

Etiology- and seizure type-specific incidence

Idiopathic/cryptogenic seizures were most common, accounting for 54.6% of incident cases (Fig. 3). Among remote symptomatic and progressive symptomatic cases, cerebrovascular disease predominated (18.8%) followed by head injury (7.3%) and brain neoplasm (6.8%). The incidence of idiopathic/cryptogenic seizures was greater than the incidence of remote symptomatic seizures, which in turn was greater than the incidence of progressive symptomatic seizures (Table 1).

Figure 3.

Distribution of etiology for all unprovoked seizures.

The adjusted incidence was 16.6/100,000 for generalized seizures and 23.0/100,000 for partial seizures (Table 1).

Cumulative incidence

The overall cumulative incidence of unprovoked seizure and epilepsy was 1.4% (0.9–2.0%). The cumulative incidence was similar for non-Hispanics (1.4%, 95% CI = 0.6–2.1%) and Hispanics (1.4%, 95% CI = 0.5–2.3%). The cumulative incidence was also comparable for non-Hispanic blacks (1.5%, 95% CI = 0.3–2.7%) and non-Hispanic whites (1.2%, 95% CI = 0.1–2.3%, Fig. 4).

Figure 4.

Cumulative risk of single unprovoked seizure and epilepsy by race-ethnicity over the 2-year study period.


There were a total of 32 deaths among the participants from June 8, 2003 to May 25, 2007; a median follow-up of 2.9 years. More than half (62.5%) of the deaths occurred within 1 year of seizure onset. The cumulative probability of mortality over the follow-up was 17% (95% CI = 12–24%). Females had a significantly higher risk of mortality (23%, 95% CI = 16%-34%) compared to males (13%, 95% CI = 7–23%; log rank p = 0.014). Cases with both remote symptomatic (32%, 95% CI = 22–45%) and progressive symptomatic (38%, 95% CI = 18–69%) etiologies experienced significantly higher mortality compared to those with idiopathic/cryptogenic etiology (4%, 95% CI = 1–13%; log rank p < 0.001). The cumulative probability of mortality was similar across groups defined by ethnicity, seizure type, and household income.

Among cases with newly diagnosed epilepsy and first unprovoked seizures, the SMR was 1.7 (95% CI = 1.1–2.3, Table 3). SMRs were increased for males and females. The SMR was increased in cases under 65 years of age and in those aged 65 and older (Table 3). SMRs were significantly increased in groups defined by Hispanic ethnicity, partial seizure type, middle and high income, and for remote symptomatic etiology. Only the groups with idiopathic and progressive symptomatic etiologies, low income, and non-Hispanic ethnicity did not have a significantly increased SMR. Similar trends in the SMR were observed using the Washington Heights and Inwood death rate as the standard.

Table 3.  Standardized mortality ratios using the 2004 Washington Heights and Inwood death rates and 2003 U.S. death rates as the standard
 Observed deathsExpected deathsSMR (95%CI)
US, 2003 as standard
Total3219.21.7 (1.1–2.3)
 Male11 6.71.6 (0.8–2.8)
 Female2112.61.7 (1.0–2.5)
Age, yrs.
 Under 6512 1.58.0 (4.1–13.2)
 65 and over2017.71.1 (0.7–1.7)
 Hispanic17 7.22.4 (1.4–3.6)
 Non-Hispanic1510.91.4 (0.8–2.2)
 Under $15,0001210.91.1 (0.6–1.8)
 $15,000–49,00013 5.92.2 (1.2–3.6)
 $50,000 and over 7 2.33.0 (1.2–5.7)
Seizure type
 Partial2111.91.8 (1.1–2.6)
 Generalized 9 7.01.3 (0.6–2.3)
 Idiopathic/cryptogenic 3 3.40.9 (0.2–2.2)
 Remote symptomatic23 9.82.3 (1.5–3.4)
 Progressive symptomatic 5 5.90.8 (0.3–1.8)


The overall age- and sex-adjusted incidence for all medically identified first unprovoked seizures or newly diagnosed epilepsy in Northern Manhattan was 41.1 per 100,000 person-years. The incidence of a single unprovoked seizure was 23.9 per 100,000 person-years and the incidence of epilepsy was 16.4 per 100,000 person-years. These rates are somewhat lower than those reported in similarly designed population-based studies. Age-adjusted incidence of all unprovoked seizures have ranged from 48 (Annegers et al., 1999) to 69 (Hauser et al., 1993) per 100,000 person-years. The lower incidence rate in our study is unlikely to be due to a true lower incidence of unprovoked seizure and epilepsy in the community, but rather to underascertainment of a subgroup of cases. Germane to this point, the proportion with idiopathic/cryptogenic seizures was lower than expected (54.6% vs. 70% expected) (Hauser et al., 1993; Annegers et al., 1999), especially among non-Hispanics (42%) and those with higher income (40%), particularly in working aged people. If the lower incidence that we observed of idiopathic unprovoked seizure in high-income non-Hispanics were due to an increased frequency of symptomatic etiologies in this subgroup compared with other populations, we would expect to see an overall increase in the incidence rate in this subgroup. However, we found incidence rates that are lower than we expected relative to other studies, so it is unlikely that the lower than expected proportion of idiopathic-unprovoked seizures is due to a higher than expected frequency of symptomatic cases. Thus, it is possible that there was differential access to care in Northern Manhattan with some working aged non-Hispanic community residents of higher income seeking care in private physician offices outside the community. Consequently, they would not enter our sample.

Our findings are consistent with prior studies (Hauser et al., 1993; Kotsopoulos et al., 2002) in that we observed a higher age-adjusted incidence of unprovoked seizure in males (46.6 per 100,000 person-years) compared to females (35.9 per 100,000 person-years). Our finding of 54.6% with partial seizure type is also consistent with past studies. Previous population-based research has suggested that partial seizure type makes up anywhere from 20% (Oun et al., 2003) to 66% (Cavazzuti, 1980) of incident seizures.

Similar to a study conducted in Texas between 1988 and 1994 (Annegers et al., 1999), we found no difference in incidence by ethnicity. However, incidence of epilepsy may have been underestimated in the study by Annegers because workers and their families were insured under the HMO and such individuals are usually healthier than the general population. Annegers was also unable to accurately assess incidence in the elderly due to a small HMO enrollment among patients over the age of 65 years.

Few other studies have examined the incidence of epilepsy by race and ethnicity in diverse populations within single geopolitical units. Instead, most studies of incidence have been restricted to specific racial or ethnic groups, or nationalities in which specific ethnicities are not examined separately (Lavados et al., 1992; Rwiza et al., 1992; Tekle-Haimanot et al., 1997). Using comparisons of such international populations to infer an effect of race or ethnicity per se is generally not useful due to likely confounding effects of culture, environment, and varying economic and healthcare resources.

With regards to socioeconomic status (SES), our findings on incidence by household income groups are consistent with the association between low SES and increased risk for epilepsy reported in studies from England (Heaney et al., 2002) and Iceland (Hesdorffer et al., 2005). In Iceland, however, this association was restricted to adults with idiopathic/cryptogenic etiology.

The number of observed deaths in our study exceeded that expected in the US population. We also found that the cumulative probability of mortality was greater in females than in males and greater in cases with remote symptomatic or progressive symptomatic etiology versus idiopathic cryptogenic etiology. These results are similar to those reported in previous incidence studies (Forsgren et al., 2005). Consistent with prior research, we observed increased SMRs with respect to age and remote symptomatic etiology (Cockerell et al., 1997; Lhatoo et al., 2001; Forsgren et al., 2005). The SMRs were high in our study, consistent with other studies, which generally find higher SMRs in the first years after diagnosis, after which they decrease substantially (Hauser et al., 1980; Shackleton et al., 1999; Lindsten and Forsgren, 2000; Lhatoo et al., 2001). The SMR was also increased with respect to Hispanic ethnicity, a finding that deserves further attention in future studies.

Our study has several strengths. First, we used a combination of case ascertainment methods in different settings to identify our cohort. Second, our study was population-based in an economically disadvantaged community of diverse racial and ethnic background. Third, the study was not restricted to medical record review but encompassed detailed interviews in a large subgroup.

As discussed previously, one limitation of our study was that we had to rely on medical records in order to obtain the race-ethnicity of cases who were not interviewed. However, we were able to conduct a sensitivity analysis, comparing assignment of race-ethnicity using medical records and to that using interviews, and found that the medical record information was valid. Also, because our noninterviewed cases had missing data on household income, we imputed the missing data with respect to the known distribution of median household income within Northern Manhattan. This was the most conservative approach, because the distribution of low income in the interviewed cases was greater than in the source population. Additionally, our results are consistent with previous research showing an increased risk for epilepsy with decreasing household income (Heaney et al., 2002; Hesdorffer et al., 2005).

Our findings indicate little difference in epilepsy incidence across groups defined by race and ethnicity in a poor urban community. Although lower income was associated with higher incidence, higher income was associated with an increased SMR. Future research should examine reasons for differential incidence by income.


This work was supported by a grant from the Centers for Disease Control and Prevention (MM-0322). The authors would like to thank the study participants, as well as all those who worked on the study but do not qualify for authorship: Martha Orbe, MD, Nora Hernandez, MD, and Carrie Mills, MPH.

Conflicts of interest: We confirm that we have read the Journal's position on issues involved in ethical publication and affirm that this report is consistent with those guidelines. None of the authors has any conflict of interest to disclose.

Disclaimer: The findings and conclusions in this report are those of the authors and do not necessarily represent the views of the Centers for Disease Control and Prevention.