Analysis of hospitalization expenditure and its inuencing factors for lung cancer patients in China: a research based on the “System of Health Accounts 2011”

Background At present, lung cancer is the common malignant tumor of respiratory system worldwide. The disease poses a serious health threat and a substantial economic burden to patients. The economic risks brought by lung cancer arouses widespread public concern. Therefore, the target of this research is to calculate lung cancer expenditure in China under the framework of the System of Health Account 2011 (SHA 2011). Intensive studies on working mechanism of inuencing factors to lung cancer expenditure will be further explored, in order to achieve the goal of controlling lung cancer expenditure. Methods A multistage stratied sampling method was conducted in Liaoning province in China, and a total of 23559 patients were included into hospitalization expenditure analysis according to the framework of SHA 2011. The relation between the total hospitalization expenditure of lung cancer and its inuencing factors including of demographic characteristics, diagnosis and treatment, and hospital condition was analyzed with the multiple stepwise regression analysis. The impact mechanism of these inuencing factors was revealed through path analysis and survival analysis. Results The total hospitalization expenditure of lung cancer was $ 1581.8. Multiple stepwise regression analysis indicated that the total hospitalization expenditure was associated with length of stay, surgery, hospital level, insurance status, and hospital type, according to the sequence of standardized estimate (β). Length of stay contributed the most to the model R-square. Path analysis showed that surgery, hospital type, and insurance status not only made a direct impact on the hospitalization expenditure, but also made an indirect impact on it through the length of stay. Through survival analysis, we found self-funded patients of lung cancer were quicker to run out of the affordable money. Conclusions Lung cancer brought a heavy economic burden for patients. More ecient and stringent clinical control strategies should be conducted to limit the increase of the expenditure.

With the occurrence of lung cancer, it brings a heavy economic burden to the society and family. The overall direct costs of caring for patients of lung cancer in Europe were amounting to more than €3 billion per year [3]. The developed countries, like England, Germany, France and Switzerland, the average period of patients' overall disease course was 19.8 months, with the direct medical cost of €10,000-30,000 per person [4]. Overall expenditures for treating early non-small cell lung cancer (NSCLC) ranged from $ 24,040 to $97,774 (Canadian Dollar) in 13 countries [5]. In China, a systematic review investigating lung cancer cost in Chinese population from database inception to September, 2017 through PubMed, Scopus, the Cochrane Library, Chinese National Knowledge Infrastructure (CNKI), Wan Fang Data, and VIP Data was conducted, and found the medical cost for the inpatient per-time in one year ranged from ¥8,582 to ¥65,562 [6].
On a global scope, although the researches on hospitalization expenditure and in uencing factors of lung cancer have already been carried out, current studies on lung cancer expenditure lack the support of systematic accounting framework, decreasing the accuracy of expenditure calculation result and the comparability among different countries. In this research, the System of Health Accounts 2011 (SHA 2011) framework is introduced. SHA 2011 is revised and promoted by international organizations like OECD, EUROSTAT and WHO, the purpose of which is to establish a comprehensive and internal uni ed accounting system that can be compared across nations. SHA 2011 separates current health expenditure and capital formation from total health care expenditure. Three core dimensions, namely nancing schemes, health care provider, and health care function are included. The key information in expanding dimensions like disease categories, age, sex, and region of health care consumers are included [7][8].
Therefore, under the framework of SHA 2011, we could conduct institutionalizing collection and analysis of the data of lung cancer, and were also able to accurately extract medical expenditure including drug, laboratory tests, surgery, etc, as well as the patients' demographic and clinical information like medical insurance, hospital type, etc.
Theses information could also provide a chance to analyze the key points associated with cost control. Therefore, the linear connection between in uencing factors and lung cancer expenditure was explored in this research. We hypothesized that the mediating effect also existed, thus the further path analysis and survival analysis were carried out to analyze the indirect impact on the lung cancer expenditure of the in uencing factors through intermediate variables.
Nowadays, China is facing the double burden of prevention and treatment for infectious disease and chronic disease. As one of the main chronic diseases, lung cancer not only threatens people's health, but also brings heavy economic loss to the whole society. On the premise of ensuring the quality for diagnosis and treatment, with scienti c use of medical resources and effective control on treatment cost for lung cancer, the economic burden will be signi cantly reduced for the patients' family and the society. We hope that this research could systematically reveal the economic risks of lung cancer, meanwhile provides key points to the government and medical institutions for the prevention and treatment of lung cancer, and realizes reasonable intervention on the high treatment cost of lung cancer.

Data source
According to the framework of SHA 2011, the expenditure data was collected from inpatients and outpatients of the sample medical institutions in Liaoning province, from 2015 to 2017. Liaoning province is located in northeast of China, with 43.59 million population. It is one of the provinces with the highest incidence rate of lung cancer in China. The incidence of lung cancer in Liaoning Province in 2015 was 78.5 / 100,000, which was higher than the incidence of lung cancer in the whole country [9].

Study sample
A multistage strati ed sampling method was conducted in the eld survey. The rst stage was to select the sample areas. A total of 5 cities including Dalian, Fushun, Jinzhou, Tieling, and Panjin were selected, according to economic development standard and health service provision, etc. Then, 1 district and 2 counties were randomly selected in the above ve cities. The second stage was to select the medical institutions. A total of 1227 medical institutions, including 83 public health institutions, 83 hospitals, 1061 grass-roots health care institutions, were selected, according to the type and level of institutions.
Patients' information was collected to establish a database in these sample institutions, including age, sex, admission and discharge date, disease, International Classi cation of Disease, the 10th revision (ICD-10), total expenditure, cost items, and insurance type, etc. The database included 5. The cost data of lung cancer was extracted from the database. Patients with ICD-10 code 'C34' were selected. Further screening was conducted, based on the following inclusion criteria: 1) inpatients in the institution type of hospital, 2) age ≥ 18 years old. Through data cleaning of extreme and missing values, repeated data, logic errors, a total of 23559 valid records were included into nal analysis.

Assessment of hospitalization expenditure for patients
The hospitalization expenditures and reimbursements for patients from the year 2015 to 2017 were calculated. This calculation included the total expenditure and nine speci c categories: 1) test, 2) check, 3) diagnostics, 4) treatment, 5) surgery, 6) drugs 7) nursing, 8) bed fee, and 9) other (medical consumables, etc).
Measurements of demographic characteristics, diagnosis and treatment, and hospital condition Demographic characteristics included age, sex, and insurance status. Insurance status was categorized as urban employees' basic medical insurance (UEBMI), urban residents' basic medical insurance (URBMI) and new rural cooperative medical system (NCMS), and self-funded, according to the sequence of reimbursement level. Diagnosis and treatment consisted of length of stay and surgery. Hospital condition was assessed based on 2 items: 1) hospital level, and 2) hospital type. Hospitals could be divided into provincial hospital (PH), municipal hospital(MH), and district and county hospital (DCH) according to its level, while provincial level representing the highest standard of hospitals in this area. Judging from hospital type, hospitals could be assessed into three unordered categories, namely general hospital (GH), special hospital (SH), and traditional Chinese medicine hospital (TCMH). Multiple regression model was conducted to analyze the impact on the total hospitalization expenditure of these in uencing factors.
Analysis of the impact mechanism of the factors Subsequent path analysis was implemented to explore the direct and indirect impact on the hospitalization expenditure for all the in uencing factors. We established the path analysis diagram using multiple regression model, based on the following assumption.
Furthermore, survival analysis was conducted to quantify the indirect impact on the expenditure of the in uencing factors through length of stay, if assumption was proved. Patients met the criterion of end event were included. End event was de ned as 'the catastrophic health payment occurred'. We studied how long it would last in different factor groups before end event occurred, and established survival plots. Catastrophic health payment was calculated according to the explanation recommended by WHO-'household health expenditures account for 40% or more of non-food consumption expenditures'.

Statistical analysis
The difference in uencing factor groups in the univariate analysis was tested by the Wilcoxon rank sum test. A multiple stepwise regression analysis was used to clarify the in uencing factors for the total hospitalization expenditure and length of stay. Variables signi cant at the 0.05 level in the univariate analysis were included in the model, and items with P > 0.10 were eliminated. Dummy variables were constructed for the categorical variables such as insurance status, hospital level and hospital type.
Moreover, Kaplan-Meier method was used to establish the Survival Analysis plot to explore the impact degree of these in uencing factors. In the study, the logarithmic transition was conducted, because the total expenditure, nine speci c cost items, and the length of stay were not normally distributed. The cost data were adjusted according to consumer price index (CPI), and the currency exchange rate used was US$1 to ¥ 6.9364. SPSS 21.0 (SPSS Inc., Chicago, IL, USA) was used for the statistical analyses.

Results
Basic situation and composition of the hospitalization expenditure The total median expenditure for patients of lung cancer was $ 1581.8, and that value was $1831.6, $1862.9, and $1165.1, respectively during the year of 2015 to 2017. The out-of-pocket payment (OOP) for patients decreased from $782.4 to $411.5 in the three years ( Table 1). As to the composition of the hospitalization expenditure, the proportion of drug expenditure was the highest among all the nine speci c categories (Fig. 1) In uencing factors of the hospitalization expenditure The characteristics and distributions of the total expenditure of lung cancer were shown in Table 2 (Table 3 and Table 4). Collinearity diagnostics including of the method of eigenvalue, condition index, and variance proportion was conducted to analyze the multicollinearity among the variables and showed weak multiple collinearity among them.  Impact mechanism of the factors Path analysis showed the direct and indirect impact on the hospitalization expenditure for the variables.
Surgery, hospital type, and insurance status not only made a direct impact on the hospitalization expenditure, but also made an indirect impact on it through the length of stay (Fig. 3).
Furthermore, survival analysis was conducted to quantify the indirect impact on the expenditure of the three in uencing factors (surgery, hospital type, and insurance status) through length of stay. Catastrophic health payment was $720, and we extended this value from $710 to $ 730. Thus, the OOP of 150 patients matched this criterion when they were discharged. The results of survival analysis were shown in Fig. 4. Self-funded patients were quicker to run out of the affordable money than patients with UEBMI or URBMI &NCMS (Log Rank χ 2 = 28.232, P < 0.001). In addition, patients undergoing surgery could let the money maintain much more time, compared with patients with no surgery (Log Rank χ 2 = 2.028, P = 0.154). As to the hospital type, patients in the general hospital were prone to reach the catastrophic health payment in a short time, but patients left sustained their money much longer (Log Rank χ 2 = 0.467, P = 0.792).

Discussion
This study found that the median hospitalization expenditure of lung cancer patients was $1581.8, higher than the level of $1281.7per capita for all the hospitalized patients in China [11]. Compared with the national per capita disposable income, the lung cancer patients' hospitalization expenditure occupied 32.4% of urban residents per capita disposable income, and 91.7% of the rural residents [12]. This shows that lung cancer will bring heavy burden to the entire society.
From the perspective of patients, the OOP decreased from $782.4 to $411.5 during the year of 2015 to 2017, which remarkably reduced the economic burden of lung cancer patients. However, it should be noted that among the study population, 61.8% of the patients were above 60 years old, retired with no income and subsisted on pension. In addition, the cost of treatment for malignant cancer is high. If complications like infection occurred during the course of treatment, patients' conditions would be aggravated and additional economic burden be increased [13]. Thus, the risk of illness and poverty for these patients still exist.
From the constituents of the medical cost, treatment component was the main part of the total hospitalization expenditure, and drug cost occupied a large proportion, up to 41.6%. On one hand, high priced chemotherapy drugs and molecular target drugs are applied to lung cancer patients [14][15]. The high cost is also associated with lacking of domestic alternative medicines. On the other hand, China is entering a critical period of new round health care system reform, and compensation mechanism reform is an important part. The nancial support to public hospitals by the government is relatively insu cient and public hospitals are running under pressure to a certain extent. Therefore, besides price factors, the compensation mechanism needs further improved. The problem of how to ensure all the medical institutions and physicians optimized the use of drugs is also needed to be solved. Currently, combined with other nations' reform experience, the Chinese government is proceeding a series of measure of reform. Zero-tariff policy on imported anti-cancer drugs and government's policy on pharmaceuticals purchase by public bidding are all included [16][17]. More anti-cancer drugs are brought into the medical insurance catalogue, as well as strengthening nancial investment to health care institutions and implementing health technology assessment on drugs utilization.
The research showed that hospitalization expenditure was in uenced by hospital length of stay, surgery, hospital type, hospital level and medical insurance situation. Hospital length of stay had the greatest impact on hospitalization expenditure and largest contribution to model R-Square. Since the severity of the disease, long time of observation and treatment for lung cancer hospitalized patients was needed, thereby the charges for beds, treatment and drugs increased accordingly. Moreover, for lacking of scienti c and effective management on patients, some medical institutions increased the hospital stay and the cost in the process of treatment. The results of path analysis indicated that patients hospitalized cost was directly in uenced by the stay length and indirectly by surgery, medical insurance situation and hospital type. The hospital length of stay was longer for those patients that had operations, medical insurance or having hospitalization in the special hospitals like oncology specialty hospitals. One reason was that compared with non-operative patients, the time for operative patients to recover was longer. For the patients with medical insurance, the usage of hospitalization service was increased to some extent compared with self-paid patients. The condition of lung cancer patients transferred from traditional Chinese medicine hospitals and general hospitals to specialized oncology hospitals was more severe.
Therefore, reducing the hospital length of stay was the prime way to lower down patients' hospitalization expenditure. General control strategy should be established. Through clinical pathway management strategy, lung cancer patients could be scienti cally and effectively treated and ineffective hospital length of stay be shortened, which had been demonstrated [18]. For various in uencing factors, speci c measures should be established to control the hospital length of stay. Through strict surgical indications and day surgery, hospital length of stay could be reduced [19]. By implementing diagnosis related groupsprospective payment system (DRGs-PPS) [20], the reform in UEBMI, URBMI and NCMS had been carried out. Meanwhile, hospitals' ability in lung cancer patients' diagnosis and treatment could be improved through coordination and cooperation.
In addition to hospital length of stay, patients whether or not taking operation was another direct in uencing factor to hospitalized expenditure. The reason was that patients had to pay for the anesthesia drugs and the application of ECG monitoring and respirator cost, besides the operation cost. If complications like postoperative bleeding and atelectasis happened, additional cost would be paid.
Therefore, physicians had to carefully choose the treatment plan. Currently, the treatment plan of lung cancer patients was based on pathological type and stage, chemotherapy, radiotherapy and operation were included [21]. Physicians had to select the most suitable treatment plan to patients based on their conditions.
Multiple linear regression showed that patients with medical insurance would pay higher hospitalized cost compared with the self-paid patients. Although the medical insurance system had provided more choices for patients, it should be aware that the moral hazard would appear and lead to overuse of medical resources between physicians and patients because of medical insurance [22]. Those with UEBMI had higher hospitalized cost than the patients with URBMI and NCMS. The reason was that China is still in the initial stage of developing basic medical insurance system. There was still a gap in nancing level, the scope of insurance, the ratio of reimbursement among UEBMI, URBMI and NCMS. The lung cancer patients with UEBMI would have better medical guarantee and more choices of treatment plan. As to the self-paid patients, since the limitation of economy, some patients wouldn't go to the hospital until the disease advanced, therefore missed the best time to treat. Meanwhile, they were also faced with limited options concerning the treatment cost. The results of survival analysis showed that the hospitalized cost of self-paid patients was spent quicker and the risk of catastrophic expenditure was easier to appear. It indicated that the supervision of self-paid patients' hospitalized cost should be strengthened and more attentions should be paid from medical aid.
From the aspect of hospital type, lung cancer patients spent lower hospitalized cost in traditional Chinese medicine hospital. The reason was that traditional Chinese medicine and treatment could improve the physical function of the patients, prevent the recrudescence and reduce the post-operative complications.
During the period of tumor chemotherapy and radiotherapy, traditional Chinese medicine could reduce its toxic side effects and enhance the therapeutic effect [23]. The introduction and combination of traditional Chinese medicine could make health care providers take patients as the foremost, focus on patients' life quality. It could help patients to set up an accurate view of health and life, learn to weight the advantage and disadvantage of the treatment and avoid the pain brought by excessive medical treatment [24].
Judging from the hospital level, lung cancer patients' hospitalization expenditure of district and county hospitals was obviously lower than the provincial and municipal hospitals. The charging rate and the complexity of patients' condition from high level hospitals was one important reason for the high hospitalized cost. Compared with the patients of acute symptom and the critical patients in the ICU, the price elasticity of demand existed in chronic disease such as lung cancer [25]. Price lever was useful in patients ow and reducing the burden of lung cancer patients hospitalized cost. The family physician system and hierarchical medical treatment system should be enhanced for patients' division. However, it required the lower level hospitals having strong support capacity and physicians meeting requirement of treatment. Strengthening the construction of regional medical center and hospital groups like the Britain's hospital trust was the main way to upgrade the lower level hospitals capability.
There are two limitations in this study. First, more variables needed to be included. The data of this research was collected from sample medical institutions of Liaoning province, according to the framework of SHA 2011. Since the variables were limited, some variables like the patients' income and education level were not included, which would be the in uencing factors of patients' hospitalization expenditure. Second, cross-sectional study was used in survival analysis and prospective research design was not included in the analysis of hospital length of stay for the patients reaching catastrophic expenditure rate. The patients whose self-paid cost reaching catastrophic expenditure rate but not discharge from hospitals were not included in the research and thereby reduced the sample size.

Conclusions
In the present study, we found lung cancer brought a heavy economic burden to patients under the framework of SHA 2011, especially for the patients in provincial and special hospitals, patients in more severe condition such as surgery and long hospitalization, and those without medical insurance, indicated that the government should pay more attention to the in uencing factors and their interactions, conduct more e cient ways in order to relieve the economic risk for patients of lung cancer in China. The Ethics Committee of China Medical University approved the study protocol. This analysis used secondary data, and the authors did not require direct contact with the participants, thus consent to participate was waived.

Consent for publication
Not applicable Availability of data and material The researchers who want to use the datasets in this study may contact the corresponding author for reasonable data request.

Competing interests
The authors have declared that no competing interests exist.    Survival analysis plot