Source of data
This study used the reimbursement claims data of the Taiwan National Health Insurance Program, which was reformed from 13 prior health insurance schemes and started in March 1995. With universal coverage, >99% of 22.6 million Taiwan residents have been enrolled in this system. The Taiwan National Health Research Institute collects information on all beneficiaries in the National Health Insurance Research Database (NHIRD), including patient demographics, primary and secondary diagnoses of disease, procedures, prescriptions, and medical expenditures. The institute also records all inpatient and outpatient use of medical services for all beneficiaries. To protect patients’ privacy, the electronic database was decoded with patient identifications scrambled for further public access. The study was evaluated and approved by the NHIRD research committee.
By using NHI procedure codes (http://www.nhi.gov.tw/), we examined medical claims and identified 13,103 surgical patients from among 2,010,412 persons who underwent inpatient major surgeries (defined as surgeries requiring general, epidural, or spinal anesthesia and hospitalized for more than 1 day) from 2004–2007 and had a diagnosis of epilepsy, which was defined by the International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM) code 345. To increase the likelihood of capturing true positive cases with epilepsy, at least two claims coded as epilepsy in their primary diagnosis during the 24-month period before the index surgical admission were required. For each surgical patient with epilepsy, we randomly selected four surgical patients matched for sex and age who had no preoperative history of epilepsy. Patients receiving epilepsy surgeries were excluded from this study.
Preoperative major coexisting medical illnesses were recorded from medical claims for a 24-month period preoperatively; these were acute myocardial infarction, acute renal failure, chronic obstructive pulmonary disease, congestive heart failure, diabetes mellitus, hypertension, peripheral vascular disease, mental disorders, traumatic brain injury, and stroke. Similarly to epilepsy, comorbidities were defined as those preoperative conditions coded as primary diagnosis for at least two claims of ambulatory and/or inpatient medical care during the 24-month preoperative period. A history of receiving renal dialysis was considered as preoperative functional status. Eight postoperative conditions were examined: acute myocardial infarction, acute renal failure, deep wound infection, pneumonia, postoperative bleeding, pulmonary embolism, septicemia, and stroke (Khuri et al., 2005; Ghaferi et al., 2009). Postoperative complications were defined as those postoperative conditions occurring within the 30-day periods after the index surgeries, with special limitation to exclude the presence of these conditions in the 24-month preoperative periods. These complications and subsequent overall in-hospital mortality within 30 days after index surgery were the study’s primary outcomes. We defined comorbidities and postoperative complications as acute myocardial infarction (ICD-9-CM 410), acute renal failure (ICD-9-CM 584), chronic obstructive pulmonary disease (ICD-9-CM 490–496), congestive heart failure (ICD-9-CM 428), deep wound infection (ICD-9-CM 958), diabetes mellitus (ICD-9-CM 250), hypertension (ICD-9-CM 401–405), peripheral vascular disease (ICD-9-CM 443), pneumonia (ICD-9-CM 480–486), postoperative bleeding (ICD-9-CM 998.0, 998.1, and 998.2), pulmonary embolism (ICD-9-CM 415), septicemia (ICD-9-CM 038 and 998.5), mental disorders (ICD-9-CM 290–319), traumatic brain injury (ICD-9-CM 800–804 and 850–854), and stroke (ICD-9-CM 430–438).
Also considered as parameters of medical resources use were whether the surgery was performed in a teaching hospital, types of surgery, length of stay, and use of ICU and in-hospital medical expenditures. Types of surgery included skin, breast, musculoskeletal, respiratory, cardiovascular, eye, digestive, kidney, ureter, bladder, delivery, cesarean section, abortion, neurosurgery, and others. In-hospital medical expenditure is the total medical costs estimated by the sum of all cost components including the costs of diagnosis, ward admission, examination, radiation therapy, medical treatment, surgery, rehabilitation, blood transfusion, anesthesia, medical material, medication, psychiatric specific treatment, injection, and personal copayments. Population densities (persons/km2) calculated by dividing the population (persons) by the area (km2) for each administrative unit of Taiwan were categorized into four quartiles as low, moderately, highly, and very highly urbanized areas (Shih et al., 2010). Low income status was defined as patients qualified for waiving medical copayments, a qualification certified by the Bureau of Taiwan National Health Insurance.
To explore more clinical risk factors predicting postoperative adverse outcomes for preoperative epilepsy, previous histories including hospitalization and emergency visits for epilepsy care within the 24-month period before the index surgery were also considered for further analysis.
Using chi-square tests, descriptive analyses concerning distribution of demographic status, coexisting medical conditions, types of surgery, use of inpatient medical resources (including length of hospital stay, percentage of postoperative ICU admission, and in-hospital medical expenditures), and postoperative complication and mortality rates between surgical patients with and without epilepsy were compared.
Odds ratios (ORs) with 95% confidence intervals (CIs) for 30-day postoperative mortality, and complications between patients with and without epilepsy were analyzed using multivariate logistic regression to adjust for operation in teaching hospital or not, low-income status, receiving dialysis or not, urbanization, preoperative coexisting medical conditions, types of surgery, and ICU stay. Each coexisting disease was considered as a single variable entered into the logistic regression model separately. Whether postoperative complications were associated with the epilepsy was further analyzed by preoperative diagnosis and history of emergency visits or hospitalization due to epilepsy and controlled other covariates with multivariate logistic regression. SAS software version 9.1 (SAS Institute Inc., Cary, NC, U.S.A.) was used for data analyses with two-sided probability, and a p-value <0.05 was considered statistically significant.