Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 461
B IOM Data-Gathering Effort T he Institute of Medicine (IOM) committee requested that several health systems (Henry Ford Health System, Geisinger Health System, and Veterans Health Administration) and one state records linkage system (South Carolina) gather data in response to a list of surveillance questions for their populations and analyze the strengths and limitations of their systems in generating information about epilepsy. Researchers in each system generously responded to the committee’s request and provided candid evaluations of their systems’ ability to capture data on epilepsy. The following questions were posed to each system: 1. Overall Description: What are the major features of your data system and the major ways your organization makes use of the data? • Major sources of data (billing, medical charts, surveys, vital re- cords, etc.) • Methods for identifying and classifying people with epilepsy • Capacity to follow individuals over time • Used for management, clinical, policy decision making, research, etc. • Algorithms and characterizations used • Strengths and limitations of your type of data system to report data on epilepsy 461

OCR for page 461
462 EPILEPSY ACROSS THE SPECTRUM 2. Incidence and Prevalence: • What are the overall incidence of epilepsy in your population per 100,000 person-years and prevalence per 1,000 persons? • What are the incidence and prevalence by gender, race/ethnicity, age ranges ( 64), and/or insurance status (public, private, none)? (Use Office of Management and Budget [OMB] classification for race/ethnicity, collapsing American Indian/Alaska Native, Native Hawaiian-Pacific Islander, and “two or more” into an “other” category to produce the following groups: Hispanic, non-Hispanic black/African American, non-Hispanic white, non- Hispanic Asian, and non-Hispanic other.) • What time period is covered by these incidence, prevalence, and demographic data? • Methods—short description of methods or algorithms used to make the estimates • Strengths and limitations of your type of data system to identify incidence and prevalence and at what level of granularity 3. Comorbidities: • For those patients with prevalent epilepsy, what percentage also has comorbid conditions? • For those patients with incident epilepsy, what percentage also has preexisting comorbid conditions? • Methods—short description of methods or algorithms used to make the estimates • Strengths and limitations of your type of data system to link with comorbidities 4. Health Care Services: • For those with psychiatric comorbid conditions (e.g., depression, anxiety, bipolar disorder, schizophrenia/psychosis), how many are receiving treatment for those conditions? • What is the percentage of patients in your epilepsy population re- ceiving epilepsy care by type of provider (primary care, neurologist, epileptologist)? Provide this separately for incident and prevalent epilepsy. • What is the percentage of patients in your epilepsy population with seizure medication use (mono- versus polytherapy)? With antidepressant use? With both seizure medication and antidepres-

OCR for page 461
463 APPENDIX B sant drug use? Provide this for prevalent and incident epilepsy separately. • What are annual rates of use (percentage with use, average num- ber of services among users) and costs (average) of hospital care, emergency room care, physician services, and seizure medications for individuals with epilepsy? Provide this separately for prevalent and incident epilepsy. Provide comparable figures for the full non- epilepsy patient population as well. • How many patients annually receive neurosurgical interventions, including epilepsy surgery and neurostimulator implants? Provide this separately for incident and prevalent epilepsy. • How many patients annually receive electroencephalograph (EEG), magnetic resonance imaging (MRI), or video-EEG monitoring re- lated to their epilepsy? Provide this separately for incident and prevalent epilepsy. • Methods—short description of methods or algorithms used to make the estimates • Strengths and limitations of your type of data system to assess services 5. Ideas for improving epilepsy surveillance through the use of health systems data (optional) The systems were also provided with the relevant International Clas- sification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) codes and algorithms to identify epilepsy cases, health care service use, and comorbidities: • Incident epilepsy: A single medical encounter with an ICD-9 code of 345.xx in the absence of a prior 345.xx code in the medical re- cord or two or more medical encounters on separate days each with an ICD-9 code of 780.39 in the absence of a prior 780.39 code or 345.xx code in the medical record or a single medical encounter with an ICD-9 code of 780.39 and a seizure medication prescribed for outpatient use for 3 or more months without a prior 780.39 code or 345.xx code. • Prevalent epilepsy: A single medical encounter with an ICD-9 code of 345.xx or two or more medical encounters on separate days each with an ICD-9 code of 780.39 or a single medical encounter with an ICD-9 code of 780.39 and a seizure medication prescribed for outpatient use for 3 or more months. These codes can be in the primary field or a secondary field.

OCR for page 461
464 EPILEPSY ACROSS THE SPECTRUM • Incident and prevalent cases in estimating health care service use: The health care use of prevalent and incident cases should be included, even if the incident case contributes only a day to the prevalent year. • Diagnostic fields for comorbidities: Use both the primary and the secondary diagnosis field. Mental Disorders—290-319 inclusive Other Major Neurological Disorders –Cerebral palsy—343.x –Cerebrovascular accident • 434.xx Occlusion of cerebral arteries • 435.x Transient cerebral ischemia –Dementia • 290.xx Dementias • 294.1x Dementia in conditions classified elsewhere –Parkinson’s disease—332.x –Multiple sclerosis—340 Traumatic Brain Injury (TBI) –310 Specific nonpsychotic mental disorders due to brain damage –850-854 (concussion and other) Autism—299.x Other Chronic Disease –410-414 (ischemic heart disease) –401-405 (hypertensive heart disease) Asthma—493.xx The following summaries of each system’s data-gathering effort help to identify the opportunities and barriers to surveillance of the epilepsies using linked electronic health records (EHRs). Although the data are not comparable due to the variety of methodologies used across the systems, each summary is informative about current U.S. surveillance capabilities and opportunities for improving surveillance of the epilepsies.

OCR for page 461
465 APPENDIX B HENRY FORD HEALTH SYSTEM David R. Nerenz, Ph.D. Gregory L. Barkley, M.D. Marianna Spanaki-Varelas, M.D., Ph.D. Aida Li Organizational Context The Henry Ford Health System is a large, vertically integrated system with 6 hospitals, a 1,000-member multispecialty group practice, more than 2,000 other affiliated private practice physicians, more than 30 ambulatory care centers, a 500,000-member managed care plan, free-standing emer- gency rooms, and many other components or “business units.” The Henry Ford Comprehensive Epilepsy Program at Henry Ford Hos- pital (HFH) and Henry Ford West Bloomfield Hospital (HFWBH) serves as a tertiary referral center for epilepsy care for southeast Michigan (metropol- itan Detroit) and, to some extent, for a wider area that includes the rest of the State of Michigan and northern Ohio. Some patients with epilepsy are seen as one-time consults, some are seen for ongoing care through referrals from non–Henry Ford physicians, and some are seen as part of a broader medical care relationship that includes primary care and other types of specialty care within the Henry Ford Medical Group (HFMG). Patients with epilepsy who are members of Health Alliance Plan (HAP—the system- affiliated health plan) may elect to receive care from HFMG physicians but may also elect to receive care from other physician networks. In analyzing patterns of care for patients with epilepsy then, it is a chal- lenge to distinguish visits that represent the first contact with an HFMG physician for long-standing epilepsy from visits that represent the true onset of the condition. It is also a challenge to estimate overall service use (e.g., hospitalizations, emergency department [ED] visits), since not all services are necessarily provided within the HFH-HFWBH-HFMG network. For these reasons, some analyses reported here were conducted within a defined population of individuals who were HAP members assigned to the HFMG for care; others were conducted in a larger population of patients receiving epilepsy care at the HFH, HFWBH, or HFMG who were not necessarily HAP members. Because HAP has a record of all paid claims, including claims from other hospitals or physician networks, it is possible to get a complete picture of services provided to HAP members; it is not possible to guarantee a complete picture of services provided to patients with other types of insurance.

OCR for page 461
466 EPILEPSY ACROSS THE SPECTRUM Methods HAP-HFMG Patients Using existing administrative data, we identified all individuals who were HAP members assigned to the HFMG for care for the years 2006- 2010. (This is a well-defined population used as a denominator population for a variety of research and quality improvement projects.) Using the HFHS Corporate Data Store (an administrative database with data on all inpatient and outpatient care in the HFH and HFMG used for a combina- tion of financial analysis, quality improvement, and research purposes), we identified all individuals with one or more encounters with a primary or secondary diagnostic code of epilepsy or seizure.1 For all of these individu- als, we conducted a “look-back” search in records of prior years (poten- tially as far back as 1995 for patients whose records went back that far) to identify whether there had been previous inpatient or outpatient encounters for epilepsy. If no, cases were then labeled as “incident cases” for the year in which the first coded encounter occurred. If yes, cases were labeled as “prevalent cases” in any year in which an epilepsy-related encounter oc- curred. Incident cases in any one year typically became prevalent cases in later years, but patients with encounters in only one year were counted as incident cases in that year and were not counted as prevalent cases. Patients with All Insurance Types Using the Corporate Data Store, we identified all patients who had had one or more inpatient or outpatient encounters for epilepsy or seizure dis- order (using the same ICD-9 diagnostic codes) at the HFH or with HFMG physicians in 2009 or 2010. We then conducted look-back analyses for these patients to identify the first coded encounter at the HFH or HFMG for epilepsy, the site of care for that first encounter (e.g., clinic, hospital, ED), and the specialty department of the first encounter. Sample for Full Medical Record Review Because of concerns about limitations of the administrative data, we created a random sample of cases that had been identified in the HAP- HFMG cohort of both incident and prevalent cases. We conducted a fo- cused review of the complete electronic medical record (EMR) for these 1 ICD-9 codes to identify epilepsy: 345.0, 345.00, 345.01, 345.1, 345.10, 345.11, 345.2, 345.3, 345.4, 345.40, 345.41, 345.5, 345.50, 345.51, 345.6, 345.60, 345.61, 345.7, 345.70, 345.71, 345.8, 345.80, 345.81, 345.9, 345.90, 345.91, 780.39.

OCR for page 461
467 APPENDIX B patients to confirm diagnosis of epilepsy, use of anti-epileptic medications, and use of antidepressant medications. Incidence or Prevalence Incidence estimates were calculated for each of the 5 years 2006-2010, using the number of incident cases (definition above) as the numerator and the number of HAP-HFMG-assigned individuals in each year as the denominator. Similarly, prevalence estimates were calculated each year and then again for the entire 5-year period by identifying the unique patients included in any one year as the numerator and the unique individuals who were in the denominator populations in any year as the 5-year denominator. Patient Demographics Patient age, gender, and race or ethnicity were available as standard data elements in the Corporate Data Store. Patient age was recorded in the year in which he or she was identified as either an incident or a prevalent case (HAP-HFMG cohort) or the year in which he or she was first seen in the 2009-2010 cohort. Use of Medications Pharmacy claims data in the Corporate Data Store for the HAP-HFMG cohort were used to identify filled prescriptions for either anti-epileptic medications2 or antidepressant medications. The claims data include pre- scriptions filled at Henry Ford pharmacies as well as “outside” pharmacies, but do not include prescriptions paid either by patients themselves or by other insurance. 2 Acetazolamide, carbamazepine, carbamazepine XR, Carbatrol, Celontin, Depacon, Depak- ene, Depakote, Depakote ER, Depakote Sprinkle, Diamox Sequels, Dilantin, Dilantin-125, di- valproex sodium, divalproex sodium ER, Epitol, Equetro, ethosuximide, Fanatrex, felbamate, Felbatol, fosphenytoin sodium, gabapentin, Gabitril, Gralise, Keppra, Keppra XR, Lamictal, Lamictal (Blue), Lamictal (Green), Lamictal (Orange), Lamictal ODT, Lamictal ODT (Blue), Lamictal ODT (Green), Lamictal ODT (Orange), Lamictal XR, Lamictal XR (Blue), Lamictal XR (Green), Lamictal XR (Orange), lamotrigine, levetiracetam, Lyrica, Mebaral, Mysoline, Nembutal Sodium, Neurontin, oxcarbazepine, Peganone, pentobarbital sodium, phenobarbi- tal, Phenytek, phenytoin, phenytoin sodium, potassium bromide, primidone, Sabril, Stavzor, Tegretol, Tegretol XR, Topamax, Topiragen, topiramate, Trileptal, valproate sodium, valproic acid, Vimpat, Zarontin, Zonegran, zonisamide.

OCR for page 461
468 EPILEPSY ACROSS THE SPECTRUM Service Utilization The Corporate Data Store was used to identify outpatient visits, ED visits, hospitalizations, or other forms of service use for epilepsy. ICD-9 diagnostic codes were used to identify epilepsy-related encounters. Current Procedural Terminology (CPT) and ICD-9 procedure codes were used to identify epilepsy surgeries and services in the inpatient Epilepsy Monitor- ing Unit (EMU). The EMU includes video-EEG monitoring for all cases, and an MRI is standard practice, either just before or just after the EMU admission. Other Patterns of Care Issues Provider, department, and site codes available for every encounter in the Corporate Data Store were used to calculate time intervals between initial presentation for epilepsy and consult with a neurologist and “flow patterns” between the ED, other sites of care (e.g., primary care), and neurology. Results Analysis of Administrative Database on an Enrolled Population Incidence or prevalence The incidence of epilepsy in the population was estimated at 266 per 100,000 in 2006 and 163 per 100,000 in 2010. There was a gradual, steady decline in estimated incidence of new cases over the 5-year study period. This incidence is considerably higher than the 48 per 100,000 reported by Hirtz and colleagues (2007). We believe that the higher incidence estimate here may reflect the fact that health plan members are free to choose a provider network and that plan members with epilepsy, or with newly diagnosed epilepsy, would be inclined to select the HFMG network upon either joining the health plan or receiving the diagnosis. They would appear to be incident cases in our administrative data set, but some would not in fact be incident cases and others would be, but would be “self-selecting” into both numerator and denominator populations used to calculate incidence. The prevalence of epilepsy was relatively stable over the 5-year period, with each individual year yielding an estimate of approximately 4 cases per 1,000 in the denominator population. We also identified all of the individuals who had been in the denominator population in any of the 5 years studied and calculated a prevalence estimate in that larger group. The numerator in this estimate included any individual who had had an en- counter coded as epilepsy or seizure disorder at any time during the 5-year period. This prevalence estimate was approximately 8 per 1,000 (1,884 out

OCR for page 461
469 APPENDIX B of 231,347). We believe that the difference between the prevalence estimate based on single-year data and the estimate based on 5-year data reflects the fact that many patients with stable, well-controlled epilepsy are seen at in- tervals greater than one year, so they appear in the numerator once or twice in the data set in a 5-year period, but do not appear in each individual year, even though they are consistently in the denominator population. Demographics About two-thirds of both incident and prevalent cases were adults between the ages of 19 and 64. The remaining cases were evenly split between children (< 19) and older adults (65+). There were approximately equal numbers of males and females among both incident and prevalent cases. The race or ethnicity distribution of the incident and prevalent cases reflected the distribution of both health plan membership and the Detroit area, with relatively large black and non-Hispanic white groups (each approximately 40-50 percent of the total) and much smaller Hispanic, Asian, or other groups. Comorbidity Patients with epilepsy in our population also had other medical and psychiatric conditions for which they receive care. In the 1,603 incident cases for example, 1,213, or 76 percent, had at least one other coded diagnosis at an HFMG medical encounter. In the 3,258 cases who had either incident or prevalent epilepsy, 1,174, or 36 percent, had another psychiatric condition coded for at least one visit, along with epilepsy. Sources of care Virtually all patients had at least one physician encounter of some kind in any one study year. The average number of physician office visits for incident cases in the year in which they were diagnosed was ap- proximately 12; the average number of physician office visits for prevalent cases in any year in which they had at least one visit at all was in the range of 9-10. Most encounters for which epilepsy was coded were with neurolo- gists. Fewer than 20 percent of cases have a recorded ED visit (although ED visits at hospitals outside the Henry Ford system would not be recorded); 25-30 percent of cases have visits with primary care physicians, and ap- proximately 75 percent have at least one visit with a neurologist. Use of medications The pharmacy claims data for both incident and prevalent cases did not show any filled prescriptions at all for 20 percent of the patients. Although this could conceivably reflect a true absence of prescriptions filled, it seemed to us more likely that to be a reflection of pa- tients’ having drugs paid for through an insured spouse or perhaps having a benefits plan with a high deductible for prescription drugs so that some prescriptions were not shown as having been paid for by HAP. Keeping this issue in mind, we found that 25-30 percent of the incident

OCR for page 461
470 EPILEPSY ACROSS THE SPECTRUM cases in any one of the 5 years had a filled prescription for anti-epileptic medications in that year and 55-65 percent of the prevalent cases had a prescription for anti-epileptic medications in any one of the 5 years. Ap- proximately 20 percent of both incident and prevalent cases had a prescrip- tion for antidepressant drugs in any one of the 5 years. Approximately 5-10 percent of the incident cases and 15 percent of the prevalent cases had both types of medications in any one year. Because all of these pro- portions seemed unreasonably low, we generated a random sample of 100 cases from the lists of both incident and prevalent cases in order to more carefully analyze the use of prescription drugs by doing a complete review of the patients’ EMRs. Medical Record Review Of the 100 cases selected for full medical record review, 72 were con- firmed as having epilepsy, either through text in physician notes or text from EEG or EMU reports; 6 of the remaining 28 had possible epilepsy, but the diagnosis either was not confirmed by EEG testing (e.g., patient was seen in the ED several times and did not return for EEG evaluation) or was in some other way ambiguous. Of the 22 remaining patients, the primary reasons for reactive seizures other than epilepsy were encephalopathy, brain tumor, alcohol withdrawal, or hydrocephalus. In one case, a neurocardio- genic syncope was the diagnosis eventually given to what had originally been labeled as a seizure. All but one of the 72 cases with confirmed epilepsy were receiving seizure medications. That one patient had been seizure-free since 1989 and seizure-free after having been weaned off anti-epileptic medications for 2 years prior to the 5-year study period. Use of antidepressant medications was much less common in these patients; only 7 of the 72 confirmed cases were prescribed antidepressant medications during the 5-year study period. Administrative Data on Hospitalizations and ED Visits The proportion of patients hospitalized in any one year was higher among incident cases than among prevalent cases, perhaps reflecting ad- missions to the EMU as part of the process of establishing epilepsy as a diagnosis for seizures. The mean number of hospitalizations for a patient in any one year was in the range of 1.7-2.2 for both incident and prevalent cases, among those with any hospitalizations at all. The maximum number of hospitalizations observed in any one year was 13 for incident cases and 22 for prevalent cases. The proportion of incident cases with at least one hospitalization in each year ranged from 43 percent in 2006 to 55 percent in 2010. The proportion of prevalent cases with at least one hospitalization

OCR for page 461
471 APPENDIX B in each year was stable in the range of 26-29 percent across the 5 years studied. ED care was relatively stable in its occurrence, both across study years and in incident and prevalent cases. Among incident cases, the proportion with at least one ED visit ranged from 30 to 38 percent in specific study years. Among prevalent cases, the range was 29 to 33 percent. There were on average of two to three ED visits per year among those patients who had any ED visits at all, among both incident and prevalent cases. (We note that not all ED visits were for epilepsy or epilepsy-related problems.) Surgical treatment was relatively rare. There were only seven surgeries among 1,603 incident cases in the 5-year study period and 24 among the 1,884 prevalent cases. This rate is, however, higher than that reported na- tionally. Our higher rate probably reflects the presence of a well-respected epilepsy surgery program in the medical group and the potential for health plan members who might be candidates for surgery to elect the HFMG network and thereby enter both numerator and denominator of the surgery rate. Patterns of Care for Patients with All Insurance Types There were 9,588 patients in 2009-2010 who met criteria for epilepsy based on ICD-9 diagnostic code criteria and were seen by HFMG physi- cians at one of 35 clinic sites. An additional 2,588 patients in the same time period were classified as “possible epilepsy” based on the presence of just one epilepsy code (suggesting its use as a “rule-out” diagnosis) or an ICD-9 code such as “seizure or seizure disorder” that could signify either epilepsy or some other form of seizure. The distributions of age, gender, and race or ethnicity were essentially the same in this larger sample of patients as in the cohort of HAP-HFMG patients described above. Most of the patients were in the 19-64 age range, most were either non-Hispanic black or white, and there were approxi- mately equal numbers of males and females. The proportion of patients insured by Medicare was larger than the proportion of patients over age 65, suggesting that many patients with epilepsy had obtained Medicare coverage on the basis of disability. A preliminary examination of patterns of visits to different types of providers suggested the presence of four distinct groups of patients under care for epilepsy at Henry Ford. These include the following: 1. patients in the system with a primary care relationship who develop epilepsy; 2. patients who come to the neurology department from outside the system for outpatient consult or referral;

OCR for page 461
498 EPILEPSY ACROSS THE SPECTRUM VISIT CODED AS 780.39 (Seizure not otherwise specified) >1 previous visit/year Less likely YES coded for 780.39 to be 780.39 and/or 345.xx NO >6 months and <6 years, Consider Less likely YES and concurrent illness 780.31 to be 780.39 that could cause fever NO For observations Seizure medication Less likely with chart reviews YES listed in chart or to be 780.39 seizure drug level done NO CPT coded for vagus nerve Less likely YES stimulator implantation to be 780.39 (95970 and 95974) NO CPT coded for genetic Less likely YES testing of epilepsy to be 780.39 (83891-83912) NO Coded for epilepsy Consider Do not code YES surgery (6153x) 345.xx 780.39 NO Code as 780.39 FIGURE B-2 Decision algorithm for individuals coded with a seizure not otherwise specified (780.39). NOTE: CPT = Current Procedural Terminology.

OCR for page 461
499 APPENDIX B are the availability of a central data repository for multiple data systems in an agency that has legal authority to identify people with epilepsy and the availability of the UID linked to personal information files to contact patients as needed. Usefulness The SCESS has demonstrable usefulness for case management and service delivery, policy development, and research. Examples include iden- tifying low-income, severe cases of TBI-related epilepsy for service delivery in the Department of Disabilities and Special Needs. In this collaborative work, data gleaned from the SCESS inform resource planning based on periodic prevalence estimates and prioritizing for services. In the areas of policy development, the SCESS provided the information needed to build the case for a joint resolution (Act No. 168) to develop a comprehensive service delivery system for people with epilepsy. This act is currently pend- ing the signature of the governor. In areas of clinical services, planning is under way to incorporate epilepsy care in underserved communities via tele- medicine platforms. The overwhelming evidence of need for this approach emanated from the surveillance information. Data show that 40.7 percent of people with epilepsy in the state reside in rural counties that require at least a day’s trip to see a neurologist. In areas of research, the SCESS continues to be critical for development of pilot projects and cooperative grants by providing the preliminary data needed for research applications. Other uses include public information and education in an annual event known as “Epilepsy Boot Camp” and dissemination of brochures to health workers and physician offices on depression among people with epilepsy. Strengths and Limitations The SCESS has several strengths. First, it is a passive surveillance system that relies on existing data sources collected for administrative pur- poses. This makes the system cost-efficient with little or no need for data solicitation. Second, the events of epilepsy are captured from a well-defined population base, making the numerator representative of the denominator. This ensures that estimates derived are generalizable and valid. Third, data acquisition is timely, providing estimates on short- and long-term trends. Currently, 15 years of person-specific data are available on epilepsy and seizure disorders, making the system among the best sources of epilepsy data for epidemiological analysis. Fourth, the data system includes UIDs that allow linkage across multiple data platforms for service delivery, clini- cal research, and outcome studies. Capacity to link electronic surveillance data with medical charts has been particularly useful to evaluate positive

OCR for page 461
500 EPILEPSY ACROSS THE SPECTRUM predictive value, sensitivity, and coding accuracy. Fifth, the data sets include information on procedures (up to 50 CPT codes) and acute care charges. CPT codes provide substantiating information on VNS implant, epilepsy surgery, genetic testing for epilepsy, and EEG monitoring to validate the diagnosis codes of epilepsy among persons coded with 789.03. Lastly, the availability of the full range of acute care charges broken down by type of service and procedure is important for cost-related comparative effective- ness studies. Despite the aforementioned strengths, there are important limitations worth noting. First, while the data system is representative and complete for the civilian population, it does not capture cases diagnosed in federal medical facilities, specifically persons from the two VA and the five military hospitals. Given the high incidence of TBI-related epilepsy among Gulf War veterans, this is likely to contribute to underestimation of prevalence. South Carolina has an estimated 300,000 veterans whose risk for epilepsy is presumed to be higher than that of the general population. However, this limitation is a universal flaw of all public health data systems in the United States. Second, data come from administrative records designed primarily for billing third-party providers. This makes the coding of a diagnosis re- sponsive to the policies of providers and the preference for diagnosis codes that maximize reimbursement. Further, there is preference for diagnosis codes that are less likely to be denied, lead to reduced reimbursement, or put more financial burden on the patient. A plausible explanation for the preference of 789.03 over 345.x in the face of multiple visits is in part to avoid labeling patients with a diagnosis of epilepsy. Our data evalua- tion shows that 82.6 percent of cases coded as 789.03 are true epilepsies. Third, with wide variability in skill sets and diagnostic resources among hospitals, the accuracy of the fourth and fifth digit of the diagnosis codes from underresourced hospitals might be unreliable. Fourth, the CPT codes are nonspecific for assessing if all EEGs, video-EEGs, and MRIs are related to the diagnosis of epilepsy without medical record evaluation. Likewise, cost estimates for epilepsy are “contaminated” by costs incurred by other conditions unrelated to epilepsy, requiring the development of a better methodology for cost analysis. Incidence and Prevalence Estimates for 2006-2010 Brief Description of Methods for Estimating Prevalence and Incidence Cases of epilepsy are discriminated as incidence or prevalence based on their first encounter. Case ascertainment criteria are described earlier. A flag variable is constructed by counting the number of times a case with a

OCR for page 461
501 APPENDIX B UID is encountered. Cases with more than one encounters are labeled as “R” for repeat and “N” for new encounters. Because this report provides information on encounters since 2006, a few incident cases seen in the latter part of 2005 might have been labeled as new in 2006, making the estimated incidence in 2006 slightly higher—0.14 percent compared to 0.10 percent for the average of 2007-2010. The advantages and disadvantages of the system are described earlier. Nontabular Description of Incidence and Prevalence Incidence and prevalence were calculated taking the 2008 (the median year) population of the state as the standard. Population estimates were ac- quired by county and demographic characteristics from the CDC National Center for Health Statistics website (CDC, 2010). County-specific infor- mation on income and poverty level was extracted from the U.S. Census Bureau Small Area Income and Poverty Estimate (Census Bureau, 2011). Results show that the cumulative incidence of epilepsy from 2006 through 2010 is 0.5 percent, which yields an annual incidence of 0.095 percent, or 95 per 100,000 population per year. This estimate is much higher than the 39 per 100,000 per year reported from Rochester, Minnesota, for the period 1955-1984 (Annegers et al., 1995)—the only population-based study pub- lished based on complete case ascertainment criteria. This discrepancy is at- tributable to temporal variation and differences in population composition (Sander, 2003). By taking the mean age (32.2 years) of people with epilepsy in the state as the average duration of follow-up, person-year denominator was constructed to generate incidence density that can readily be converted to risk as proposed by Morgenstern and colleagues (1980). Accordingly, a probability of 0.0051 (5.1 in 1,000 S.C. residents) is estimated for new onset of epilepsy over the 5-year period of observation. Annual incidence by age group showed 0.19 percent, 0.08 percent, and 0.05 percent for 0-18, 19-64, and ≥ 65, respectively. Gender differences were minimal, with females at 0.11 percent and males 0.10 percent per an- num. Incidence was twice as high in blacks (0.16 percent) as in whites (0.08 percent). Incidence was 0.09 percent in Hispanics and 0.07 percent in other races. The most profound difference in incidence was noted among the in- surance categories. Medicaid-insured individuals had 26-fold increased risk of new onset of epilepsy compared to those with private insurance (0.398 percent per year for Medicaid and 0.015 percent per year for private). Incidence was 0.053 percent for Medicare and 0.020 for the uninsured. Comparison of ratios in reference to private insurance indicates that the incidence of new onset was 26.0, 3.5, and 1.3 times greater in Medicaid, Medicare, and uninsured, respectively.

OCR for page 461
502 EPILEPSY ACROSS THE SPECTRUM Annualized prevalence was in the same direction as the incidence. It was higher in the age group 0-18 (0.541 percent) followed by 19-64 (0.375 percent). The prevalence among older adults, age ≥ 65, was 0.242 percent. Analysis by gender showed higher prevalence in females (0.468 percent) than in males (0.328 percent). The magnitude of the difference in preva- lence among the race or ethnic groups was comparable to incidence, with ratios nearly twice as high in blacks (0.626 percent) as in whites (0.317 percent). The prevalence among Hispanics was 0.222 percent and in races grouped as “other” was 0.273 percent. Prevalence estimates also show the disproportionate burden of epilepsy borne by persons with Medicaid insurance (1.059 percent). This is nearly seven times higher than the preva- lence of people with epilepsy with private insurance (0.153 percent). The second highest prevalence was among persons with Medicare insurance. It is important to note the discrepant prevalence estimates observed in older adults (0.242 percent) and the high prevalence in persons with Medicare insurance (0.474 percent). This discrepancy is explained by Medicare eli- gibility criteria. Although all older adults are eligible for Medicare, not all Medicare eligibles are older adults. Medicare is also an entitlement program for persons with disability who qualified for Social Security Disability In- come. In the epilepsy data set analyzed for this report, 25 percent of people with epilepsy younger than age 65 have qualified for Medicare. In fact the mean age of Medicare insured was 55.9 (±17.6) and the median age was 55. Thus, Medicare insurance carries a large proportion of prevalent cases of epilepsy with disability as reflected by the higher prevalence than that observed among older adults. Comorbidities Brief Description of Methods for Estimating Comorbidities Co-occurrences of illnesses other than the primary disease of interest (epilepsy) are identified from the secondary diagnosis fields (9 in Medicaid and the SHP; 14 in the UB) in the data sets. Thirty-one comorbid conditions known to be associated with epilepsy beyond those that could be explained as chance and/or of interest to this report were identified using “arrays” and “do loops” in SAS V9.1.3. The SAS program was written in such a way that it identifies one disease at a time while ignoring the other comorbid diseases until the “do loop” exhausts all the diagnosis fields referenced in the ar- ray listing. This procedure allowed counting of more than one comorbid condition per patient. For example, 170 patients had 5 or more of the 31 conditions at the same time.

OCR for page 461
503 APPENDIX B Description of Comorbidities Among Prevalent Cases Of the prevalent cases, 39.3 percent have one or more comorbid condi- tions (i.e., 18.3 percent with two or more, 21.0 percent with one condition). Mental health comorbidities accounted for 14.9 percent, while cardiovas- cular diseases including established hypertension accounted for 21.6 per- cent. Diabetes mellitus and asthma with chronic bronchitis accounted for 7.9 percent and 8.1 percent, respectively. Based on corroborating Vcode (V15.82) and CPT code (99406-07), most of the chronic bronchitis cases appear to be associated with smoking. Substance abuse disorders (drugs and alcohol) were noted in 2,607 (4 percent) of the prevalent cases. Cogni- tive and learning difficulties were noted in 1,981 (3 percent) of the prevalent cases and appear to be associated with duration of illness based on the number of encounters with these patients. Stroke was noted in 2.5 percent of the prevalent cases, but it is uncertain whether it is temporally anteced- ent to the epilepsy or a subsequent event. Forty-three percent of stroke was noted among older adults with epilepsy. Another high-frequency comorbid- ity among prevalent cases is anemia, noted in 2,179 (3.25 percent) patients. While 59.5 percent of the prevalent cases are females, the proportion of females with anemia was 68.5 percent, suggesting the preponderance of females with epilepsy that have comorbid anemia. Other low-frequency but important comorbid illnesses include nutritional deficiency (N = 879; 1.3 percent), brain trauma (N = 272; 0.41 percent), multiple sclerosis (N = 265; 0.40 percent), and HIV/AIDS (human immunodeficiency virus/acquired im- mune deficiency syndrome) (N = 232; 0.35 percent). Description of Comorbidities Among Incident Cases Of incident cases, 16.2 percent have comorbid conditions. In contrast to the number of persons with comorbid illnesses among prevalent cases, comorbidity among incident cases is 60 percent less. The distribution of comorbid illnesses mirrors that of the prevalent cases with the difference being the counts of comorbidities. When proportions are derived from the number of cases with at least one comorbid illness (i.e., positive cases for comorbidity), significant differences exist between incident and prevalent cases. Chronic physical illnesses such as cardiovascular disease, diabe- tes, and asthma were significantly higher among prevalent cases, while the proportion of emotional and behavioral problems such as depression, mood, and anxiety disorders was significantly higher among incident cases: 48.3 percent of incident cases had emotional and behavioral problems in contrast to 37.5 percent of prevalent cases; conversely, 56.4 percent of the prevalent cases with at least one comorbidity had cardiovascular dis- ease, compared to 39.2 percent of incident cases. These differences in the distribution of comorbidities between incident and prevalent cases yield

OCR for page 461
504 EPILEPSY ACROSS THE SPECTRUM important information to estimate the residual risk of comorbid illnesses attributable to epilepsy. Health Care Services Brief Description of Methods Used to Estimate Health Care Services The UB, SHP, and Medicaid files were linked with mental health and substance abuse service files that provide information on service utilization in clinics run by various agencies. Additional information pertaining to re- ceipt of services was acquired with the CPT codes of 96150-96155, which indicate treatment for psychological, behavioral, emotional, and cognitive health problems. Information on access to specialty care was identified from rendering the specialty label included in all of the data sets utilized. Professional specialties were grouped in the following manner. Evaluations made by neurologists and neuropathologists were listed as a “neurolo- gist care”; neurological (epilepsy) surgeons as “neurosurgery”; evaluations made by neuropsychiatrists and psychiatrists as “psychiatric care.” Evalua- tions made by family physician, internist, pediatrician, emergency medicine, and general practitioner were listed as “primary care.” All other consults and evaluations made by various specialties, including radiologist, nurse practitioner, psychologist, neuropsychologist, et cetera, were grouped as “all other care.” Receipt of care for psychiatric problems was determined by the specialist rendering the service or by referral disposition to mental health clinics, which when flagged indicated that the service was received. Venues of care were grouped as inpatient, hospital outpatient, or ED; phy- sician offices; and ambulatory care services. Annual rate of use by venues of care was estimated by counting the total encounters made in each of the venues and expressed as a proportion. The algorithm used to identify epi- lepsy cases and recency of onset (incidence) is described earlier. Information on seizure medication use and most common prescription was identified from 2,226 randomly selected chart reviews in the state. The abstraction expenses were covered by funding from the CDC, NCCDPHP Epilepsy Program Office. Estimates for selected services are provided by the CPT codes. Direct cost of medical care was derived from charged amount per specialty and venues of care. According to the ORS, the charge-to-revenue ratio in South Carolina is $1.0:$0.92. Cost summary is analyzed using SAS “Proc tabulate” with “sum*$charge” and “mean*$charge” options. Information on provider specialty was missing in 24.3 percent of the cases. In these circumstances, missingness was determined to be completely at random and ignorable when comparisons of demographic, hospital, and payer characteristics of observations with missing and nonmissing values were not significantly different.

OCR for page 461
505 APPENDIX B Receipt of Care for Psychiatric Problems There is some discrepancy between the number of people with estab- lished diagnosis of psychosis, depression, mood and anxiety disorders, and receipt of care for these problems. While 9,489 (10.6 percent) of the total 89,938 people with epilepsy had the mentioned diagnoses, 7,570 (8.4 percent) received treatment. This suggests that of those with these psychiatric diagnoses, 79.8 percent received treatment for mental health problems, which included therapies offered by primary care physicians, clinical psychologists, and psychological counselors. The number of people with epilepsy who received treatment from psychiatrists was only 856 (0.93 percent of those with mental health problems). Receipt of Epilepsy Care Of the total 67,040 prevalent cases of epilepsy identified from 2006 to 2010, 22.8 percent were diagnosed and treated by neurologists (includes the 18 epileptologists in the state); 59.6 percent were evaluated and treated by PCPs; and 16.3 percent were evaluated and treated by other providers. Of the total 22,898 incident cases of epilepsy, 32.1 percent had evaluation and treatment rendered by a neurologist; 55.8 percent by PCPs; and 11.9 percent by other providers. Seizure Medication Types and Combinations Information on treatment relied on 2005 chart reviews since the surveil- lance data are not linkable to pharmacy files. Further, while revenue codes based on National Drug Codes are available, there are too many codes for the same generic product depending on dosage, routes of administration, and brand names, making such linkage unwieldy. Data from chart reviews of randomly selected 2,226 people with epilepsy showed that 70.5 percent were only on monotherapy; 24.2 percent were on two medications; and 5.3 percent were on three or more seizure medications. The most commonly prescribed seizure medications were phenytoin (55 percent), valproic acid (19 percent), carbamazepine 18 percent, phenobarbital (13 percent), and gabapentin (6 percent). Fifteen other seizure medications have usage rate of 5 percent or less. Odds of taking more than one seizure medication was influenced by severity (adjusted odds ratio = 1.72; 95 percent CI 1.29- 2.30). Unfortunately, this information was completed earlier and could not be separated by incidence and prevalence. Similarly, it was not possible to obtain data on antidepressant use alone or in combination with seizure medications.

OCR for page 461
506 EPILEPSY ACROSS THE SPECTRUM Annual Rates of Use and Direct Cost of Care Usage rates were estimated by rates of encounters. Over the 5 years, there were 1,226,479 encounters with 89,938 unduplicated patients. The average encounters per patient per annum were 2.73. The most frequently utilized venue of care was the hospital-based ED at 26.6 percent. Most of the ED encounters were made by Medicaid and uninsured people with epilepsy, suggesting the disproportionate reliance of these patients on the ED as their medical home. Medicaid accounted for 55.4 percent of the total 1,226,479 encounters contributing to the heavy utilization of the ED. Med- icaid patients have limited quota in private practices because of the very low reimbursement rate of Medicaid. Inpatient hospital care has the second- highest utilization rate per annum at 22.9 percent, with a preponderance of children and older patients for admission regardless of insurance status. Hospital-based outpatient services were the third most common venues of care, accounting for 17.8 percent of the encounters. Case mix was 33.8 percent Medicaid, 25.4 percent Medicare, and 25 percent private insur- ance. There were an average of 38,757 private physician office visits per annum accounting for 15.8 percent of the total encounters. The case mix was predominantly Medicare and Medicaid. EEG, psychological testing, imaging, and laboratory evaluation accounted for 16.9 percent of the visits. The average charged amount per annum was $6,884 for inpatient care, $586 for ED care, $469 for hospital-based outpatient care, and $186 for private office visits. Hospital-based bills, EDs, and OPDs include procedure charges that are less frequently rendered in private offices. This analysis was not able to partition total charges per service into subcharges. Receipt of Neurosurgical Interventions There were 5,173 surgical interventions over the 5 years of observation, with annual interventions averaging 1,034. For this analysis, interventions were not partitioned by procedure types. The average cost of neurosurgical intervention ranged from $1,809 for Medicaid to $5,602 for commercial insurers, with an overall average of $4,501.00 per intervention. The total charge included the whole range of neurosurgical interventions from insert- ing and replacing a neurostimulator pulse generator in outpatient surgery to lobectomy. The great majority (90 percent) of the interventions were implants. Detailed information on annual rates of use and costs of hospital care, ED care, and physician services in a given year; average number of services per setting; cost of seizure medications; and comparison to non-epilepsy population were not available. Furthermore, it is not possible to partition services by prevalence and incidence status until supplementary data ele- ments are acquired from the sources.

OCR for page 461
507 APPENDIX B Suggestions and Recommendations For population-based analysis and public health activities, systematic and ongoing surveillance of epilepsy is best conducted by using existing multifaceted data sources. It will be ideal if there is a centralized agency or organizations, such as the ORS, that has the legal authority to serve as a data repository in defined jurisdictions. It will be important that there be unique identifiers to link files across multiple data platforms to unduplicate observations and discriminate incident and prevalent cases. Access to medi- cal charts for periodic evaluations of positive predictive value, sensitivity, and accuracy of the case ascertainment criteria is of paramount importance. Because of the chronic nature of epilepsy and the stigma associated with it, epilepsy diagnosis is frequently masked with seizure unspecified, delirium, and even syncope codes. Sufficient knowledge of these cases is acquired when corroborating evidences is available from CPT codes, medication use, prior visits, and review of records. There is sufficient evidence gleaned from periodic surveillance to indicate the disproportionate burden of epilepsy in minorities and economically disadvantaged groups, rampant payer-related substandard care, and the occurrence of comorbidities among people with epilepsy that exceed the general population threshold. The increasing trends of epilepsy in the elderly and socioeconomically disadvantaged population groups suggest the plausibility of an ecological link between the disease and socioeconomic determinants. The chronic nature of epilepsy, with its major impact on quality of life, economic impact on the national health care cost, and potential to prevent secondary conditions associated with it, are strong public health rationales supporting the need to maintain four to six sentinel sites across the nation for ongoing surveillance of epilepsy. REFERENCES Annegers, J. F. 2004. Epilepsy. In Neuroepidemiology: From principles to practice, edited by L. M. Nelson, C. M. Tanner, S. K. V. D. Eeden, and V. M. McGuire. New York: Oxford University Press. Pp. 303-318. Annegers, J. F., W. A. Hauser, J. R. J. Lee, and W. A. Rocca. 1995. Incidence of acute symp- tomatic seizures in Rochester, Minnesota, 1935-1984. Epilepsia 36(4):327-333. Berg, A. T., S. F. Berkovic, M. J. Brodie, J. Buchhalter, J. H. Cross, W. van Emde Boas, J. Engel, J. French, T. A. Glauser, G. W. Mathern, S. L. Moshe, D. Nordli, P. Plouin, and I. E. Scheffer. 2010. Revised terminology and concepts for organization of seizures and epilepsies: Report of the ILAE Commission on Classification and Terminology, 2005- 2009. Epilepsia 51(4):676-685. Berlowitz, D. R., and M. J. V. Pugh. 2007. Pharmacoepidemiology in community-dwelling elderly taking antiepileptic drugs. International Review of Neurobiology 81:153-163. CDC (Centers for Disease Control and Prevention). 2010. Bridged-race resident population estimates: United States, state and county for the years 1990-2009. http://wonder.cdc. gov/wonder/help/bridged-race.html (accessed January 6, 2012).

OCR for page 461
508 EPILEPSY ACROSS THE SPECTRUM Census Bureau. 2011. Small area income and poverty estimates. http://www.census.gov/did/ www/saipe/ (accessed January 6, 2012). Hauser, W. A., J. F. Annegers, and L. T. Kurland. 1993. Incidence of epilepsy and unprovoked seizures in Rochester, Minnesota: 1935-1984. Epilepsia 34(3):453-468. Hirtz, D., D. J. Thurman, K. Gwinn-Hardy, M. Mohamed, A. R. Chaudhuri, and R. Zalutsky. 2007. How common are the “common” neurologic disorders? Neurology 68(5):326-337. Hope, O. A., J. E. Zeber, N. R. Kressin, B. G. Bokhour, A. C. Vancott, J. A. Cramer, M. E. Amuan, J. E. Knoefel, and M. J. Pugh. 2009. New-onset geriatric epilepsy care: Race, setting of diagnosis, and choice of antiepileptic drug. Epilepsia 50(5):1085-1093. Kotsopoulos, I. A., T. van Merode, F. G. Kessels, M. C. de Krom, and J. A. Knottnerus. 2002. Systematic review and meta-analysis of incidence studies of epilepsy and unprovoked seizures. Epilepsia 43(11):1402-1409. Morgenstern, H., D. G. Kleinbaum, and L. L. Kupper. 1980. Measures of disease incidence used in epidemiologic research. International Journal of Epidemiology 9(1):97-104. Ngugi, A. K., S. M. Kariuki, C. Bottomley, I. Kleinschmidt, J. W. Sander, and C. R. Newton. 2011. Incidence of epilepsy: A systematic review and meta-analysis. Neurology 77(10):1005-1012. Pugh, M. J. V., A. C. Van Cott, J. A. Cramer, J. E. Knoefel, M. E. Amuan, J. Tabares, R. E. Ramsay, D. R. Berlowitz, and the Treatment in Geriatric Epilepsy Research Team. 2008. Trends in antiepileptic drug prescribing for older patients with new-onset epilepsy: 2000- 2004. Neurology 70(22 Pt. 2):2171-2178. Sander, J. W. 2003. The epidemiology of epilepsy revisited. Current Opinion in Neurology 16(2):165-170. Selassie, A. W., B. B. Wannamaker, P. L. Ferguson, E. E. Pickelsimer, G. Smith, and R. Turner. 2005. Final report of the South Carolina epidemiological studies of epilepsy and seizure disorders. Charleston, SC: Medical University of South Carolina. Sillanpää, M., and S. Shinnar. 2010. Long-term mortality in childhood-onset epilepsy. New England Journal of Medicine 363(26):2522-2529. South Carolina Budget and Control Board Office of Research and Demographics. 2012. Health utilization menu. http://ors.sc.gov/hd/utilization.php (accessed January 6, 2012). Thurman, D. J., E. Beghi, C. E. Begley, A. T. Berg, J. R. Buchhalter, D. Ding, D. C. Hesdorffer, W. A. Hauser, L. Kazis, R. Kobau, B. Kroner, D. Labiner, K. Liow, G. Logroscino, M. T. Medina, C. R. Newton, K. Parko, A. Paschal, P.-M. Preux, J. W. Sander, A. Selassie, W. Theodore, T. Tomson, S. Wiebe, and the ILAE Commission on Epidemiology. 2011. Stan- dards for epidemiologic studies and surveillance of epilepsy. Epilepsia 52(Suppl. 7):2-26. Weinstein, A., A. Bengier, M. A. Eccher, and F. G. Gilliam. 2011. Enzyme-inducing anti- convulsant effects on lipid levels: Absence of evidence from a patient database. Poster accepted for presentation at the American Epilepsy Society Annual Meeting, Baltimore, Maryland (December). http://www.aesnet.org/files/dmfile/PosterSession1_2011.pdf (ac- cessed January 6, 2012).