Rare diseases in China: analysis of 2014–2015 hospitalization summary reports for 281 rare diseases from 96 tertiary hospitals

In the first step of creating the TRDL 2017, rare disease names were summarized according to four available lists of rare disease names in China. These four sources included recommendations for the rare disease name list made by experts of the National Health Commission of the People’s Republic of China that was meant to improve ICD coding and funding reimbursement of therapies, experts from the Beijing Society of Rare Diseases for epidemiologic surveillance, the book entitled Treatable Rare Diseases [10] for scientific popularization of meteorites, and a national study on a partial registry of rare diseases (the National Key Research and Development Program of China clinical cohort study of rare diseases (2016YFC0901500)) that was a national fund project for rare disease research.

In the next step, after removing duplicate names, we obtained a primary list with 344 rare diseases by summarizing and proofreading disease names from the four list sources mentioned above.

In the third step, two expert consensus meetings were held. In the first meeting, 18 experts from across China were invited to individually explain their rationale for the primary list as well as the methodology involved, via public discussions. The professional fields of the 18 experts included pediatrics, neurology, respiratory medicine, ophthalmology, genetics, pharmacy, epidemiology, statistics, mathematics, and information science. In the second consensus meeting, another group of 21 experts first held public discussions and then voted by anonymous ballot for those diseases with the highest research priorities. The final TRDL 2017 was formulated based on the results of this expert consensus. The experts who took part in the two expert consensus meetings were all senior experts on relevant rare diseases nationwide. The flowchart of development of the TRDL 2017 is shown in Fig. 1.

Data were extracted from the database of hospitalization summary reports (HSRs). This is a patient-level national database of hospitalized populations. The selected hospitals submit HSRs to the HSR system annually, in accordance with requirements of the National Health Commission of the People’s Republic of China [11‐14]. The HSR system includes data integration, data storage and management, data analysis and mining, and results display. Each layer guarantees data safety and quality control [15].

The database covers 96 tertiary hospitals in 25 provinces across China. All 96 hospitals are university affiliated hospitals or provincial hospitals. For each patient in the HSRs database, clinical information includes demographic characteristics (age, sex), discharge diagnosis, location of the hospital, and corresponding ICD-10 codes.

Target rare diseases in the TRDL 2017 were identified according to discharge ICD-10 codes. The flowchart of data capture is shown in Fig. 1.

Demographic information about the study population and their admissions to tertiary hospitals during 2014 to 2015 in China, including the number of hospitalizations, male to female ratio, city distribution, age distribution and readmission rate.

Rare diseases were analyzed by their ICD-10 codes. Correctly identifying disease names in the HSRs database is complex as the database contains English names, names in both English and Chinese, transliteration of Chinese names, and synonyms. To minimize possible inaccuracy of disease coding and names, both ICD-10 codes and key search terms (in English and Chinese) of rare diseases were used for data capture. In addition, a few rare diseases lacking ICD-10 codes were identified using search terms (in English and Chinese). The total number of hospitalizations, total cases of rare diseases on TRDL 2017, the top ten rare diseases with most cases and the rare diseases with no more than one case were calculated.

Patients’ information on the residential province of patients could not be obtained; therefore, hospital locations were used for city distribution. The five cities with the most types and the five cities with the most cases of rare disease listed on the TRDL 2017 were calculated.

Patients’ age at admission was used for analysis of age distribution. The age group included 0–4 years, 5–14 years, 15–24 years, 25–34 years, 35–44 years, 45–54 years, 55–64 years, 65–74 years, 75–84 years, 85~ years. The number of these ten age groups were calculated.

Hospitalizations of patients in the same hospital could be identified, but not in different hopitals due to the deidentification and encryption of patient data. So readmission in this study refers to rehospitalization in the same hospital.

Continuous data were described using mean and standard deviation; and categorical variables were presented as frequency and proportion. All statistical analyses were performed using R (version 3.5.1).

The present study is the first national survey of rare diseases in China and included the largest study population to date. Second, the process from development of the TRDL 2017 to data capture and analysis was rigorous. Third, based on a systematic methodology, we established the TRDL 2017 is a feasible way, and the list can be continuously and quickly updated for further study. Finally, our study will contribute to updating the World Health Organization nomenclature for rare diseases in China in that we standardized the names of 281 rare diseases between English and Chinese language.

Although this hospitalized population-based study could describe the fundamental data of a sizable group of rare diseases, underreporting of rare disease cases is inevitable for three reasons. First, the HSR data are limited to hospitalized patients. Second, all hospitals involved in this study are all tertiary hospitals, but not all tertiary hospitals in China were included in the database. Third, tertiary hospitals in China also provide primary, secondary and tertiary care and have the exposure to nationwide patient population due to the lack of hierarchical referral system, which was different from the tertiary hospitals of western medical system, so prevalence in each city could not be obtained. Fourth, by cross-matching our TRDL list to Orphanet nomenclature of RD, we found that most diseases in our list are single diseases, and some are groups of diseases, which may lack precise ICD-10 codes so could not be extracted from the database. Fifth, mismatching of rare disease nomenclature may have resulted in the exclusion of some patients. Sixth, the statistical results of the research data are biased caused by the fact that the current registration information of inpatients in different hospitals in China cannot be shared so the hospitalization number of same patient with rare disease admitted to different hospitals cannot be offset. For example, rehospitalization rate was underestimated as the rehospitalized cases only represent inpatient cases in the same hospital. Consequently, individual-level data could not be acquired in the present study. Seventh, residential place of the hospitalized patients is not an essential parameter in the database. Therefore, the distribution of the patient population by city are unclear. Lastly, the final selection of the 281 diseases on the TRDL 2017 was determined by anonymous ballot as those diseases with considered to have the highest research priority, which makes this list very different from those of other publications focusing on disease frequency. However, our results still fill a gap in the data for rare diseases in China. It is the largest and most complete dataset with important reference value.