The data of the first 4 cohort datasets (R01-R04) in 2001 from the National Health Insurance Research Database (NHIRD; http://www.nhri.org.tw/nhird/) in Taipei were used for the analysis. The NHIRD owned all claims in electronic form from the national health insurance (NHI) program that has started in 1995 and covered nearly all inhabitants in Taiwan (21 653 555 beneficiaries at the end of 2001)[24]. The NHIRD has retrieved dozens of datasets publicly available for researchers. For each cohort dataset, the NHIRD at first randomly sampled 50 000 people from 23 753 407 people who had ever been insured from March 1, 1995 to December 31, 2000. Then, all insurance claims belonging to these people were drawn to make up one specific cohort dataset.

A cohort dataset included one registration file of the insured people and 6 files of original claim data: inpatient expenditures by admissions, details of inpatient orders, ambulatory care expenditures by visits, details of ambulatory care orders, expenditures for prescriptions dispensed at contracted pharmacies, and details of prescriptions dispensed at contracted pharmacies. The structure of the insurance claim files had been described in details on the NHIRD web site and in other published literature[25-28].

Among the 4 cohort datasets in 2001, there were totally 183 976 eligible people, 22 746 admissions, 1 241 760 inpatient prescription items, 2 607 646 visits (including 234 598 visits for traditional Chinese medicine), 11 765 537 outpatient prescription items, 127 008 dispensed prescriptions at pharmacies, and 513 231 dispensed prescription items at pharmacies. The ambulatory sector included medical care services at clinics of Western medicine, traditional Chinese medicine, and dentistry, and pharmaceutical services at pharmacies. Because the separation of prescribing and dispensing was not yet thoroughly executed in Taiwan, relatively few outpatient prescriptions had been dispensed at independent pharmacies.

Besides, the fee schedule of 4 837 medical service items and the list of 21 146 approved drug items of Western medicine in Taiwan were downloaded from the web site of the Bureau of National Health Insurance (http://www.nhi.gov.tw/, accessed January 12, 2002). Each drug of different brand, strength and form had an official unique identifier for claims. The BNHI also offered a list of ATC codes (the Anatomical Therapeutic Chemical classification system, version 2000)[29] for each drug item.

The costs of gastroenterological diseases were estimated in 4 dimensions: diagnosis, diagnostic procedure, therapeutic procedure, and drug. The reason for separate approaches was that on the one hand an admission or a visit with a gastroenterological diagnosis might contain service items not related to the gastroenterological diagnosis, and on the other hand a procedure or drug for gastroenterological diseases might not be associated with any gastroenterological diagnosis on the claim because of the coding error or limitation (5 diagnosis fields on an admission claim and 3 on a visit claim).

The age - sex prevalence of patients having any gastroenterological diagnosis during the year was at first calculated. All 5 diagnoses on an admission record and all 3 diagnoses on a visit record were taken into account. Then, medical costs of visits and admissions because of gastroenterological diseases as the principal diagnosis were calculated and stratified by major disease category.

The definition and categorization of gastroenterological diseases were based on the scheme of the Clinical Classifications Software (CCS for ICD-9-CM) developed by the Agency for Healthcare Research and Quality (AHRQ) of the U.S. Department of Health and Human Services[30]. The single-level diagnosis CCS aggregated illnesses and conditions of over 12 000 diagnosis codes within the International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM) into 259 mutually exclusive categories. The single-level CSS categories 6, 12-18, 120, 135, 138-155, 214, 250, and 251 were deemed as gastroenterological diseases. Some codes such as benign neoplasm, injury, and screening that were dispersedly included in other CSS categories were also identified as gastroenterological diseases.

The fee schedule of the NHI included mainly consultation and rehabilitation fees, charges for diagnostic procedures (e.g. laboratory tests and radiological examinations), therapeutic procedures (e.g. general treatments, radiological interventions and surgeries) and materials, and case payments based on diagnosis-related groups. Among 1 470 items of diagnostic procedures and 2 435 items of therapeutic procedures, 132 and 338 items were related to gastroenterological diseases respectively.

Drugs for treating gastroenterological diseases in the current study included all drug items of the groups A02 (antacids, drugs for treatment of peptic ulcer and flatulence), A03 (antispasmodic and anticholinergic agents and propulsives), A04 (antiemetics and antinauseants), A05 (bile and liver therapy), A06 (laxatives), A07 (antidiarrheals, intestinal antiinflammatory/antiinfective agents), A09 (digestives, including enzymes), C05A (antihemorrhoidals for topical use), P01A (agents against amoebiasis and other protozoal diseases), and P02 (anthelmintics) in the ATC classification system. The items in the group J07BC (hepatitis vaccines) were not included because they were not reimbursable within the NHI. A total of 2 269 drug items for treating gastroenterological diseases had been registered in Taiwan since 1995. Some drugs might be no more available on the market or not reimbursable by the insurance during the study period.

In describing utilization and costs of diagnostic procedures, therapeutic procedures, and drugs related to gastroenterological diseases, the number of recipients, the number of orders, and the total costs were calculated and stratified either by individual item of procedures or by group of drugs at the ATC 3rd and 4th levels. The original monetary values in unit of New Taiwan dollars were converted into U.S. currency based on the average exchange rate in 2001 (1 U.S. dollar [USD] = 33.8003 New Taiwan dollars) according to the Central Bank of China in Taiwan (http://www.cbc.gov.tw/, accessed 6 February 2003).

The database software of Microsoft SQL Server 2000 was used for data linkage and processing. The regular statistics were displayed.

Among the 200 000-person cohort, only 183 976 people were still insured in 2001. There were more men than women (92 566 vs 91 394), and the status of sex was unknown in 16 persons (Table 1). The mean age of the eligible people was 33.9 (SD 20.3) years.

The medical care benefits of these people in 2001 totaled 83 830 239 USD, of which 26 443 645 USD (31.5%) was claimed at the inpatient sector and 57 386 595 USD (68.5%) at the ambulatory sector. The clinics of traditional Chinese medicine and independent pharmacies only accounted for 4.2% (3 499 397 USD) and 0.8% (677 908 USD) of the total costs respectively.

The utilization and costs of all kinds of procedures and drugs among the cohort were as following: 1 119 629 orders of diagnostic procedures summing to 7 858 087 USD (9.4% of the total costs), 774 340 orders of therapeutic procedures with 23 582 229 USD (28.1% of the total costs), and 8 683 664 prescribed or dispensed items of drugs with 21 193 285 USD (25.3% of the total costs).

Table 1 also displays the number of patients with any gastroenterological diagnosis in 2001 and their prevalence among the eligible cohort, stratified by age group, sex, and setting. Totally, 44.2% of the insured people had at least one gastroenterological diagnosis. The age group 20-39 years had the lowest prevalence rate while the elderly had the highest. The prevalence rate was higher in women than in men (48.5% vs 40.0%). But, men had a slightly higher prevalence rate (2.9%) at the inpatient sector than women (2.0%). During the year, 14 888 patients were admitted to hospitals. In other words, 30.4% of inpatients had ever a gastroenterological diagnosis at discharge. Physicians of traditional Chinese medicine made gastroenterological diagnoses in 9 643 patients, i.e. in 5.2% of the entire cohort and in 18.8% of the 51 359 patients having visited the clinics of traditional Chinese medicine during 2001.

Table 2 shows the aggregate costs of visits and admissions due to gastroenterological diseases. The analysis took account of the principal diagnosis into consideration on each claim only, the results were stratified by major disease category, and the numbers of affected patients were also calculated. Totally, 67 073 patients (36.5% of the cohort) were involved and medical care costs in value of 8 469 909 USD (10.1% of the total medical costs) were consumed. On average, 46.0 USD per insured person in a year was spent in treating gastroenterological diseases.

Besides, 16.2% (n = 3 676) of all admissions, 8.0% (n = 209 254) of all visits, and 10.4% (n = 24 429) of all visits to traditional Chinese medicine during 2001 were primarily attributed to gastroenterological causes. While the largest patient group was noninfectious gastroenteritis (9.6% of the cohort), followed by other gastrointestinal disorders (8.7%), abdominal pain (7.7%), gastritis and duodenitis (7.7%), and other disorders of stomach and duodenum (4.2%), the top 5 diagnosis groups with higher total charges were gastroduodenal ulcer (10.6% of the costs related to gastroenterological diseases/1.1% of the total medical costs), other liver diseases (6.7%/0.7%), other gastrointestinal disorders (6.7%/0.7%), hepatitis (6.5%/0.7%), and biliary tract disease (6.0%/0.6%).

Among the cohort during 2001, 36 004 (19.6%) patients utilized 210 156 (18.8%) diagnostic procedures related to gastroenterological diseases with costs of 1 900 752 USD (24.2% of the costs for all diagnostic procedures and 2.3% of the total medical costs). The computer tomography with 3 fee items took almost a two-fifth share of the costs for GI-related diagnostic procedures. The other two items with higher costs were abdominal sonography (18.4% of the costs for GI-related diagnostic procedures) and upper GI panendoscopy (13.0%). But, the most common procedures belonged to biochemical tests of blood: glutamic-pyruvic transaminase (GPT) with 47 523 orders and glutamic-oxalacetic transaminase (GOT) with 45 653 orders. Nearly a seventh of the cohort received these two tests (Table 3).

Among the cohort during 2001, 8 349 (4.5%) patients utilized 60 245 (7.8%) therapeutic procedures related to gastroenterological diseases with costs of 1 066 497 USD (4.5% of the costs for all therapeutic procedures and 1.3% of the total medical costs). The most common general treatments were nasal feeding, insertion of nasogastric tube, and gastric decompression. The most common surgeries were hernioplasty, appendectomy, cholecystectomy, hemorrhoidectomy, and trans-arterial embolization. The utilization of surgeries was far less frequent than that of general treatments (Table 4).

Over four-fifths of the cohort received drugs for treating gastroenterological diseases in value of 1 551 250 USD (7.3% of the costs for all drugs and 1.9% of the total medical costs). The most popular drugs were antacids (ATC group A02A) prescribed to 141 185 (76.7%) patients, drugs for treatment of peptic ulcer (A02B) with 29 181 (15.9%) recipients had the largest (33.1%) share of drug costs related to gastroenterological diseases. Table 5 gives a breakdown of the utilization into the fourth level of ATC classification. Despite fewer recipients, proton pump inhibitors (A02BC) were the second in aggregate cost only to H₂-receptor antagonists (A02BA). Another noteworthy fact was that drugs for liver therapy were prescribed to 7.1% of the cohort.