Pneumonia prevention effects of perioperative oral management in approximately 25,000 patients following cancer surgery

Abstract Aim We conducted a multicenter study to explore the risk factors of developing pneumonia and the effectiveness of perioperative oral management (POM) for the prevention of pneumonia in postsurgical patients. Methods and results A survey covering eight regional hospitals was conducted over 4 years, from April 2010 to March 2014. Using the Diagnosis Procedure Combination database, a target group of 25,554 patients with cancer who underwent surgery was selected and assessed from a population of 346,563 patients without pneumonia on admission (sample population). The study compared the incidence of pneumonia and attempted to identify the significant predictive factors for its occurrence in these patients using multiple logistic regression analysis. Comparative assessment for the occurrence of pneumonia before and after POM implementation showed a significant incidence decrease after POM introduction in the target group, with no such change observed in the sample population. Multiple logistic regression analysis showed that the odds ratio for pneumonia occurrence after POM introduction was 0.44, indicating a reduced risk of pneumonia. Conclusion POM in cancer patients was indeed effective in reducing the incidence of pneumonia in hospitals and thereby helped in preventing pneumonia during hospitalization.

POM refers to professional oral care provided by dentists or dental hygienists and involves pretreatment evaluation of oral hygiene and administration of appropriate oral care and dental treatment in cases of high risk of infection. The goal of POM is to ensure good oral hygiene prior to initiating further treatment, as well as posthospitalization. Several studies have shown that perioperative oral care may help prevent complications such as pneumonia and surgical site infection after cancer surgery. (Hong, Gueiros, Fulton, et al., 2019;Ishimaru et al., 2018;Nishikawa et al., 2019;Sato et al., 2011;Soutome et al., 2017) However, as these were small-scale studies and high-evidence-level studies are limited, it remains unclear whether POM can reduce pneumonia incidence. (Aoyama & Tamagawa, 2019) The number of cancer patients is increasing; in 2016, there were 17.2 million cancer cases worldwide with an increase of 28% between 2006.(Global Burden of Disease Cancer C et al., 2018 Therefore, the number of patients hospitalized for treatment is also expected to increase. However, the number of the patients that can be treated by dentists or dental hygienists is limited and it is difficult to perform POM for all cancer patients admitted to the hospital. It is therefore necessary to consider the factors that are implicated in the onset of pneumonia and to prioritize POM for patients who are at high risk. So far, the factors that expose hospitalized patients to increased risk of developing pneumonia remain unclear. For example, no study to date has compared the incidence of pneumonia in hospitalized patients by cancer type.
Here, we planned a multicenter collaborative study on the incidence of pneumonia in cancer patients and attempted to determine which cancer patients are at increased risk of developing pneumonia.
We screened these patients for cancer types with high risk of developing pneumonia during hospitalization and compared the incidence of pneumonia before and after the introduction of POM.
POM is only provided to cancer, organ transplantation, and cardiovascular surgery patients in the Japanese universal health insurance system. Our study only targeted cancer patients who underwent surgery, and we examined how often these patients developed pneumonia, whether there were any cancers associated with the development of pneumonia other than esophageal and lung cancers, which have already been reported, (Agostini et al., 2010;Ando, Ozawa, Kitagawa, Shinozawa, & Kitajima, 2000) and whether the incidence of pneumonia would be reduced by the initiation of POM.
The purposes of this study were to evaluate the effect of POM by comparing the preintroduction and postintroduction of POM periods, to statistically calculate the incidence and risk of pneumonia by cancer type and to determine base data for triaging hospitalized patients who undergo POM.

| METHODOLOGY
This was a multicenter, collaborative, retrospective study using data from the Diagnosis Procedure Combination (DPC) database. The details of the DPC database have been described elsewhere. (Fuji, Akagi, Abe, et al., 2017;Nakamura, 2016;Okamoto et al., 2018;Okamoto, Uchiyama, Takemura, et al., 2013) The DPC data of hospitalized patients included all electronic records pertaining to clinical and medical care information.
Of all patients admitted to the participating institutions, only cancer patients who underwent surgery were selected using the DPC database.

| Variables
The variables retrieved from the DPC database were sex, age, date of hospitalization, three categories of the diseases names (encompassing diseases that led to hospitalization, comorbid diseases at the time of hospitalization, and diseases that developed during hospitalization), and DPC codes.

| Patients
Patients were recruited from eight regional central hospitals. All institutions adopted the bundled payment healthcare fee system using the DPC system.
Since POM was introduced into the Japanese universal health insurance system in April 2012, the period of analysis covered 2 years before (April 2010-March 2012) and after (April 2012-March 2014 its introduction. The sample population included 346,563 subjects, all of whom were medical patients hospitalized in any of the eight facilities during the 4-year study period, excluding those who had pneumonia at the time of admission. All cancer patients who underwent surgery, excluding those who had pneumonia at the time of admission, were extracted from the sample population as the target group, ultimately including 25,554 patients.

| Patient selection and data collection
The DPC database includes three categories of diseases, encompassing diseases that led to hospitalization, comorbid diseases at the time of hospitalization, and diseases that developed during hospitalization. For example, when a patient with hypertension is hospitalized for the purpose of treatment of stomach cancer, the disease that led to hospitalization would be registered as stomach cancer, and the comorbid disease at the time of hospitalization would be registered as hypertension. If this patient developed pneumonia during hospitalization, pneumonia would additionally be registered in the "disease that developed during hospitalization" field.
In this study, patients with no record of any type of pneumonia in the "disease that led to hospitalization" and "comorbid disease at the time of hospitalization" fields, but with record of pneumonia, irrespective of disease type, listed as a "disease that developed during hospitalization" were considered patients who developed pneumonia during hospitalization.
The patients that did not develop pneumonia during hospitalization included those with no description of pneumonia, irrespective of type, in any of the three fields described above. "Excluded cases" referred to patients with pneumonia, irrespective of type, listed in the "disease that led to hospitalization" and "comorbid disease at the time of hospitalization" fields.
It was difficult to determine using the DPC database the exact point in time when pneumonia occurred after admission; therefore, we designated any incident case of pneumonia as pneumonia during hospitalization, instead of using terms such as postoperative pneumonia.
The DPC database includes a 14-digit code called the DPC code.
This code is specific to each patient and describes the patient's cause of hospitalization, whether surgery or treatment was performed, presence of side effects, and others. The first six digits of the DPC code indicate the name of the disease; the DPC codes were used to classify the disease distribution of the sample population into 19 disease groups, and the target group was classified into 28 types of cancer. In the target group, we classified patients based on their status of surgery and extracted all cancer patients who underwent surgery from the database (Figure 1).
Because this study included all cancer patients eligible for participation, sample size calculation was not performed.
All patient selections in this study were based on the DPC database. Therefore, information that was not included in the DPC database, that is, treatment details, stage of disease, and whether or not POM was performed, could not be included in the analysis.

| Statistical analysis
Statistical analyses were performed using EZR version 1.35. (Kanda, 2013) For a general understanding of the factors contributing to the development of pneumonia, during hospitalization, age, sex, the incidence of pneumonia during hospitalization, and the category of disease (cancer) were compared between before the introduction of POM (hereinafter referred to as preintroduction) and after the introduction of POM (hereinafter referred to as postintroduction) in the sample population and the target group. Age was compared using the t-test, and sex and the incidence of pneumonia during hospitalization were compared using the chi-squared test. In the target group, the incidence of pneumonia during hospitalization by age and cancer type were calculated, and the incidence of preintroduction was compared to that of postintroduction using the chi-squared test. Moreover, multiple logistic regression analysis was performed. The objective variable for the analysis was the onset of pneumonia during hospitalization, and explanatory variables were preintroduction and postintroduction F I G U R E 1 Scheme of patient selection of POM, sex, age, and cancer type. Age was classified into five variables (younger than 50 years, 50-59 years, 60-69 years, 70-79 years, 80 years or older), and cancer type was classified into five variables (brain, stomach, esophagus, lung, and others).
In all analyses, two-tailed p values <.05 were considered statistically significant.

| Ethics
This study was conducted in accordance with the ethical guidelines for medical and health research involving human subjects, the ethical guidelines for epidemiological research of the Ministry of Health, Labor and Welfare of Japan, and with approval from the ethics committee of each facility. The details of the study and the method of refusal to participate were publicly available, and written informed consent was not required.

| Characteristics of the patients
Characteristics of the sample population and the target group are summarized in Tables 1 and 2, respectively. In the sample population, significant differences in age and sex were observed preintroduction and postintroduction. There was no significant difference preintroduction and postintroduction in the incidence of pneumonia.
In the target group, significant differences were observed preintroduction and postintroduction in age, sex, and incidence of pneumonia.
Regarding the disease distribution of the sample population, diseases of the digestive system were the most common preintroduction and postintroduction, with the number of patients increasing in the order of cardiovascular, female genitalia and puerperal, respiratory, and nervous system diseases. Regarding disease distribution in the target group, stomach cancer was the most common, followed by colon, rectum and anus, liver and intrahepatic bile duct, and lung cancer.

| Comparison of the number of patients with pneumonia during hospitalization and the incidence of pneumonia preintroduction and postintroduction
The overall population included 169,787 patients who were hospitalized over the 2-year preintroduction period and approximately 176,776 patients who were hospitalized in the 2-year postintroduction period. The overall incidence of pneumonia during hospitalization was the same (1.6%) between the preintroduction and postintroduction groups (Table 1).
In the target group, the number of cases of pneumonia preintroduction was 11,886, and the number of cases postintroduction was 13,668. The number of patients with pneumonia during hospitalization was 239 (2.0%) preintroduction and 114 (0.8%) postintroduction, demonstrating a significant difference in the incidence of pneumonia (p < .001; Table 2).

| Comparison of the incidence of patients with pneumonia by age
The incidence of pneumonia in patients aged 10 years was remarkably high (7.1%), amounting to one in every 14 patients developing pneumonia. In the subsequent age groups, however, the incidence was as low as 0-0.3% in patients aged up to 40 years. The incidence increased after 50 years of age, and it was shown that the older the age, the higher the incidence of pneumonia. In a comparison of preintroduction and postintroduction, in the age groups older than 50 years, the incidence of pneumonia was lower postintroduction, and the margin of decrease was largest in 90-99-year-old patients ( Figure 2) 3.4 | Differences in the incidence of pneumonia during hospitalization categorized by cancer type and changes preintroduction and postintroduction Preintroduction, the incidence of pneumonia during hospitalization was highest in the order of stomach, esophageal, small intestine and peritoneal cancer, and brain tumor. Postintroduction, the incidence was highest in the order of brain, small intestine and peritoneal, esophageal, gallbladder, and extrahepatic bile duct tumor.
In a comparison of preintroduction and postintroduction, the inci- Note. Pre-introduction: before the introduction of perioperative oral management. Post-introduction: after the introduction of perioperative oral management.
F I G U R E 2 Changes in the incidence of pneumonia by age during hospitalization in cancer patients who underwent surgery (target group). Comparison between before and after the introduction of perioperative oral function management A statistically significant difference was found in the incidence of stomach (p < .001) and lung (p < .05) cancers. Prostate cancers and brain tumors showed increase in incidence, but no statistical significance was observed. Besides, the incidence of preintroduction pneumonia was 0.8% in the whole target group, and incidences exceeding this value were noted, in descending order, in patients with tumors at the following sites: the brain, small intestine and peritoneum, esophagus, pancreas and spleen, lung, stomach, and colon ( Figure 3). Sex-wise, the risk in males was 2.04 times than in females (95%  It has been reported that oral care prevents pneumonia. (Akashi, Nanba, Kusumoto, & Komori, 2019;Mizuno et al., 2018;Yoneyama, Yoshida, Matsui, & Sasaki, 1999) Conversely, some treatment guidelines do not require oral care. (Lassen et al., 2009;Mangram, Horan, Pearson, Silver, & Jarvis, 1999) This discrepancy is considered to be due to the fact that there are no high-level evidence studies on the effects of POM. In order to conduct studies at a high level of evidence, one must first select cases of hospitalized patients at high risk of pneumonia and prescribe POM treatment as an experimental intervention. Such a study would elucidate the triage method for patients at high risk of developing pneumonia and the optimal pneumonia prevention protocol. Thus, it is necessary to establish a method to select hospitalized patients who are at high risk of developing pneumonia. In this study, we examined the incidence of pneumonia in hospitalized patients with various cancer types. The results of this study will help select diseases that may benefit from POM.

| Factor analysis for pneumonia during hospitalization in the target group
In our analysis, the incidence of pneumonia during hospitalization, preintroduction, was 2.0% but decreased to 0.8% postintroduction.
The logistic regression analysis indicated that the odds ratio was approximately 0.4 postintroduction compared with preintroduction and that the risk of pneumonia during hospitalization remarkably decreased.
We targeted cancer patients because they receive POM and, therefore, the effect of POM could be evaluated by comparing the incidence of pneumonia before and after the introduction of POM in this patient sample. Moreover, it is reasonable that we found no difference in the incidence of pneumonia before and after POM introduction in the overall sample population because most did not receive POM.
We consider that POM contributed to the difference in the incidence of pneumonia among cancer patients preintroduction and postintroduction. Although this is only a comparison of the period before and after introduction, as we could not examine whether POM was actually performed or not in this study, POM intervention was considered to be the independent risk factor for developing pneumonia in cancer patients who underwent surgery.
In the future, based on the results of this study, a prospective study will be conducted to evaluate the effectiveness of POM in the specified procedure and frequency.

| Comparison of the incidence of pneumonia during hospitalization based on cancer type
In this study, postintroduction, the incidence of pneumonia during hospitalization decreased in patients with cancer of the stomach, lung, esophagus, colon, head and neck, kidney, rectum, anus, and small intestine and peritoneum.
The incidence of respiratory complications after surgery is high in patients with cancer of the esophagus and lung, and the effectiveness of oral care has been reported by retrospective studies. (Akutsu et al., 2010;Iwata et al., 2019;Shin, Kunisawa, Fushimi, & Imanaka, 2019;Tanada, Hoshikawa, Sato, Takahashi, & Koseki, 2019) We validated the findings of previous reports by providing statistical analysis using large-scale data. Additionally, the present study clarified that the inci- In the logistic regression analysis, patients with brain tumors were at a higher risk of developing pneumonia than were those with malignant esophageal, stomach, and lung tumors. Pneumonia is the main postoperative complication in brain tumor, (Nagle et al., 2017) and in cerebrovascular disorders such as cerebral infarction, it has been reported that the development of pneumonia affects the length of hospitalization and prognosis.(NanZhu, Xin, Xianghua, Jun, & Min, 2019) Conversely, no study has examined the effectiveness of POM in brain tumor patients. Therefore, in the future, more detailed investigations of pneumonia after brain tumor surgery are necessary.

| The risk factors for pneumonia during hospitalization
In the target group, the age and sex compositions were different preand postintroduction. Therefore, multivariate logistic regression analysis was performed for more accurate analysis.
The investigation of related factors showed that age was a related factor because the odds ratio significantly increased after the age of 50 years.
In contrast, as shown in Figure 2, the incidence of pneumonia uniformly and remarkably decreased after the age of 50 years. Therefore, POM appears to effectively prevent pneumonia in patients older than 50 years. POM is only performed upon request by the patient's attending physician in Japan, and it may be challenging for the attending physician to determine which patients require POM. Therefore, age may be a readily accessible POM indicator.
The incidence of hospital-acquired pneumonia increases after the age of 60 years, and the rate of aspiration pneumonia has been reported to be 50% in patients in their 60s, 70% in patients in their 70s, and over 80% in patients older than 80 years. (Teramoto et al., 2008) Our data showed a trend similar to that reported by a previous study; it is consider that most cancer patients develop aspiration pneumonia after surgery. As a method of preventing aspiration pneumonia, oral care, including evaluation and training of swallowing function, brushing, and removal of teeth that that are poorly maintained, is recommended. (Mandell & Niederman, 2019) Therefore, active intervention by POM including evaluation and training of swallowing function is considered necessary for people older than 60 years.
Sex-wise, the risk of pneumonia development was almost twice as high (p < .001) in males. It has been reported that males are at a higher risk of developing pneumonia requiring hospitalization, and older men have a higher risk of developing aspiration pneumonia due to reduced swallowing function. (Lin et al., 2019) Furthermore, swallowing function training is known to effectively prevent pneumonia. (Miki et al., 2018) Therefore, it is necessary to perform POM not only for oral care but also for evaluation and training of swallowing function for the prevention of pneumonia in elderly man.

| LIMITATIONS OF THIS STUDY
This study has some research limitations. As we only examined data included in the DPC database, information that was not provided, that is, treatment details, stage of disease, and whether or not POM was performed, could not be included in this study. Thus, individual cases could not be analyzed in detail. Therefore, it was difficult to examine the relationship between the implementation of POM and the degree of progression of the original disease, the magnitude of surgical invasion, postoperative course, incidence of pneumonia during hospitalization, and other variables. In order to clarify these points, it is necessary to subject the data of the high-risk groups to more detailed analysis that will account for background factors. The present study provides preliminary data for prospective studies. Because some important background factors were validated through this study, we plan to examine the pneumonia-preventing effect of POM after controlling for these factors in the future.

| CONCLUSION
In cancer patients who had undergone surgery, the incidence of pneumonia significantly decreased after the introduction of POM, and males were twice as likely to develop pneumonia. Patients older than 60 years had a significantly higher risk of developing pneumonia than had patients younger than 50 years.