Approximately 50% of the global population is infected with Helicobacter pylori,1 which causes various gastrointestinal diseases including peptic ulcer disease and gastric cancer.23 Recently, H. pylori infection has been associated with several diseases, including hematologic, cardiovascular, and neurological diseases, in addition to gastrointestinal diseases.45 To determine the impact of H. pylori eradication therapy on these diseases, population-based studies utilizing large-scale, nationwide, secondary databases have been performed.678 The National Health Information Database of the Korean National Health Insurance Service (NHIS), which is a mandatory health insurance system that covers the entire Korean population, is one such database.9 It has an extensive dataset of 1.3 trillion records, which includes sociodemographic status, diagnoses, operations, and prescriptions.6910 It also includes follow-up data, spanning approximately 20 years.611 Studies utilizing secondary databases may help clinicians better understand the clinical characteristics and treatment outcomes of rare diseases (e.g., metachronous lesions after gastric cancer treatment). Furthermore, they may help analyze diseases requiring long-term follow-up (e.g., the risk of gastric cancer in the general population). However, these studies may be limited by the lack of well-established operational definitions. In most studies, H. pylori eradication was defined arbitrarily without validation.678 Moreover, the success or failure of H. pylori eradication therapy could not be determined because such results were unavailable in the secondary databases.678 Because post-eradication persistence of H. pylori infection is an important confounding variable in H. pylori-related diseases, it is difficult to draw conclusions without knowing the outcome of the eradication therapy. Therefore, a definitive conclusion could not be reached by exploring the secondary databases alone.

To facilitate the use of secondary databases in studies and help reach a more definitive conclusion, we developed an operational definition of H. pylori eradication therapy. We also developed and validated operational definitions for predicting successful or failed treatment based on the prescriptions of eradication confirmatory tests and rescue therapy. These operational definitions will help researchers better analyze and interpret data from secondary databases.

This is the first study to assess the operational definition of H. pylori eradication therapy. We developed an operational definition based on the major regimens utilized in the prospective nationwide registry established in 2010–2015. Minor regimens, including dual therapy (e.g., PPI + amoxicillin, PPI + clarithromycin, or PPI + metronidazole) and quinolone-containing clarithromycin-based therapy (e.g., PPI + levofloxacin + clarithromycin or PPI + moxifloxacin + clarithromycin), were excluded from the operational definition. Dual therapy is not popular in Korea because of its low eradication rate.1415 Quinolone-based therapies are not favored, except for third-line treatment (e.g., PPI + amoxicillin + quinolone), because it is also used to treat tuberculosis in Korea, which is highly prevalent. Although some regimens were not included in the operational definition, its sensitivity to detect H. pylori eradication therapy was high, even in the validation set. This finding suggests that most regimens used in real-world settings can be predicted. The high sensitivity may be attributed to the dissemination of common guidelines and the fact that antibiotics not covered by the NHIS were infrequently prescribed in the real world. All regimens recommended in the Korean guidelines, including triple, sequential, concomitant, and bismuth-based quadruple therapies, were included in the operational definition.13

In the present study, second- or third-line therapies was defined as the prescription of one of the H. pylori eradication regimens 28 days after the previous eradication therapy. The time interval of 28 days was selected to exclude duplicate prescriptions for lost medications or early changes in the eradication regimen. In our analysis, approximately 99% of the participants received rescue therapy 28 days after their initial therapy. Given that the performance of confirmatory tests for H. pylori eradication and issuance of rescue therapy prescriptions typically occur at least 3–4 weeks after treatment, the operational definition was clinically relevant and reliable.

We also set the time interval between eradication therapy and the confirmatory test as 21 days to six months, because most participants obtained a confirmatory test during this time period. Although very few participants were subjected to confirmatory tests six months after the treatment, an interval of more than six months was not considered valid as the minimum follow-up duration was six months. The time interval was limited to six months because it may be difficult to distinguish between treatment failure and reinfection if tests are conducted after a long time.

Finally, we developed an operational definition to determine success or failure of the H. pylori eradication therapy based on a confirmatory test and the prescription of rescue therapy. In the validation set, the sensitivity and specificity for determining successful H. pylori eradication with first-line treatment were sufficiently high. Thus, the results of the first-line H. pylori eradication therapy could be determined without knowing the exact results of the confirmatory test in the secondary databases. However, the specificity for determining successful H. pylori eradication therapy after second-line treatment was relatively low. This is because administration of third-line therapy is uncommon in the real-world setting. The operational definition may mistakenly classify individuals in whom second-line therapy failed and who were not administered third-line therapy as having received successful second-line therapy. Therefore, caution should be exercised when interpreting the results of the second-line therapy.

Although our study is the first to develop and validate an operational definition of H. pylori eradication therapy, it has several limitations. First, the operational definition was developed in Korea considering the local clinical settings and preferences, including the major eradication regimens. If the operational definition is applied to other countries or regions, the preferred H. pylori eradication regimens of that country/region should be considered. Second, the institutions included in the development and validation sets overlapped, indicating that the development and validation sets may depend on each other. However, the institutions included in the validation set accounted for only a small proportion of the nationwide registry, and the study periods for the development and validation sets were distinctly separated. Moreover, the impact of the bias caused by interdependency between the development and validation sets may have been minimal. Third, the operational definition of H. pylori reinfection was not validated because of the lack of participants with reinfection in the validation set. Fourth, we did not include potassium-competitive acid blockers (P-CABs) in our operational definitions. Although tegoprazan, a P-CAB, has been approved for H. pylori eradication therapy in Korea in 2021, our study included only patients who received H. pylori eradication therapy in 2019 or before. We plan to extend our operational definitions to include patients who received P-CAB-based H. pylori eradication therapy. Although the operational definition for reinfection seems to be clinically relevant, long-term follow-up data are required to validate them. Fifth, unknown treatment results could not be ignored in the validation set, as some patients could have been lost to follow-up or may not have undergone confirmatory tests in real-world practice. Operational definitions do not provide information on the success of eradication therapy in such patients. Finally, the practical use of operational definitions in secondary databases, such as the NHIS claim database, was not tested. We plan to apply the operational definition to perform various subsequent clinical studies. A modified version of the operational definition may be derived for subsequent studies.

Despite these limitations, our study provides valid and applicable operational definitions that can be used for H. pylori-related studies utilizing a secondary database. Using prescriptions of major H. pylori eradication regimens, we developed an operational definition for identifying individuals who received H. pylori eradication therapy. Additionally, the operational definition of a successful or failed eradication therapy was based on the presence of a confirmatory test and rescue therapy prescription. Our operational definitions may be useful for various studies that utilize secondary databases.