Intensive Periodontal Treatment Reduces Risk
of
Infection-Related Hospitalization in Hemodialysis
Population
A Nationwide Population-Based Cohort Study
Shih-Ting Huang, MD, Cheng-Li Lin, MSc, Tung-Min Yu, MD, Ming-Ju Wu, MD,
PhD,
and Chia-Hung Kao, MD
INTRODUCTION
I
nfections are leading causes of death among patients with end-stage renal disease (ESRD), accounting for 15% of mortalities.1 Infectious complications are the second leadingcause of hospitalization in the ESRD population.2 In addition to
infections associated with dialysis access devices, ESRD may be susceptible to nonaccess-related infections. Among them, the most important infections include respiratory infections, Infections of the central nervous system, gastrointestinal infections,
genitourinary tract infections, cellulitis, and osteomyelitis. Among them, respiratory infections are the second leading cause of infection-related deaths.1
Numerous measures have been used for infection prevention. Recommendations for vaccination, blood-borne virus
management, and environmental cleaning are valued.3,4 Despite
these prevention measures, infection hospitalization rates in the first months of dialysis were still almost equal to rates of
cardiovascular hospitalization.2 Additional modifiable determinants
of infection prevention in dialysis patients are needed to be identified and evaluated.
Periodontal diseases (PD) are prevalent in dialysis population, with prevalence rate reaching 80.6%.5 ESRD patients
secretion than did healthy controls.6 Diseases with at least
minimal evidence of an association with periodontitis include pneumonia, chronic kidney disease, metabolic syndrome, and cancer.7 Periodontitis severity was correlated with malnutrition
and inflammatory status in dialysis patients.5 Evidence showed
that PD adversely affects all-cause or cardiovascular survival in HD population.8,9
Regarding periodontal therapy, studies have reported
improvements in endothelial function and inflammation among patients with significant PD.10,11 In clinical setting, frequent
and regular dental scaling was associated with a significant decrease in infective endocarditis (IE).12 In dialysis population,
PD therapy was suggested to improve systemic inflammation, nutritional status, and erythropoietin responsiveness.13,14
However, the clinical effects of intensive PD therapy on infection prevention in dialysis patients are largely unknown. Our study investigated how intensive PD therapy affects the risks of major infections and examined the trends of outcome risks that were modified by the frequency of intensive periodontal therapy in HD population.
METHODS
Data Source
This retrospective cohort study used data extracted from the Registry of Catastrophic Illness Database (RCID) in Taiwan, a part of the National Health Insurance Research Database (NHIRD). The Bureau of National Health Insurance (BNHI) established the universal NHI program in 1995. More than 99% of the population in Taiwan, approximately 23.72 million
people, is enrolled in the program, which provides comprehensive medical coverage
(http://www.nhi.gov.tw/english/index.-aspx). The BNHI entrusted the National Health Research Institutes (NHRI) to establish and maintain the release of comprehensive NHI-related administrative claims data for research. The RCID includes all patients who satisfy the BNHI criteria for a catastrophic illness certificate. Catastrophic illnesses are defined as severe illnesses requiring advanced health
care, such as malignancies, posttransplantation status, and ESRD. Patients with a catastrophic illness certificate are exempt
from medical care copayments, and physicians review all certificate requests. The NHIRD identifies diseases on the basis of the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM). The accuracy and validity of NHIRD diagnosis codes has been documented.15 According
to the Personal Information Protection Act, all researchers must formally declare that they have no intention of violating patient privacy. This study was approved by the Institutional Review Board of China Medical University (CMU-REC-101-012).
Sampled Participants
Figure 1 shows the process of selecting participants for the study cohorts. Patients newly diagnosed with ESRD (ICD-9-CM 585) between January 1, 1998 and December 31, 2010 and who underwent dialysis for at least 3 months were identified from the RCID. Patients with prior major infectious diseases, survived fewer than 90 days after the first dialysis date, were undergoing transplantation, were younger than 20 years, or whose information was missing were excluded. Patients with PD (ICD-9-CM code 523) who underwent intensive periodontal treatment (included subgingival curettage [scaling (91004C) and root planning (91006C, 91007C, 91008C)] and periodontal flap surgery [91009B, 91010B]) in the study period were included in the treatment cohort. The comparison cohort included HD patients without diagnosis of PD. The treatment and comparison cohorts were matched at a 1:1 ratio according to a propensity score.16 The propensity score was calculated using
logistic regression to estimate the probability of periodontal assignment according to baseline variables, namely age, sex, urbanization level, monthly income (in New Taiwan dollars), Charlson comorbidity index (CCI), and comorbidities of
chronic obstructive pulmonary disease (COPD), hyperlipidemia, diabetes, hypertension, congestive heart failure, liver
cirrhosis, and dementia. The C-statistic of the logistic regression model was 0.51.
The index date was defined as the date of intensive PD
therapy for the treatment cohort and the 15th day of the same month for the comparison cohort.
Both cohorts were observed from the index date to the date of hospitalization for an infectious disease, including acute and subacute IE (ICD-9-CM codes 421.0, 421.1, and 421.9), bacteremia (ICD-9-CM code 790.7), pneumonia (ICD-9-CM codes
487.0, 486, 481, 480.8, 482, and 484), brain abscess (ICD-9-CM code 324.0), osteomyelitis (ICD-9-CM code 730), renal and perinephric abscess (ICD-9-CM code 590.2), withdrawal from the insurance system, or the end of the follow-up period (December 31, 2010).
Independent Variables
Sociodemographic and comorbidity data, with age, sex,
urbanization level, and monthly income as the covariates, were obtained from the claims data. TheNHRI urbanization categories were adopted; these comprised 7 strata, with level 1 denoting the most urbanized communities and level 7 denoting the least
urbanized communities. Factors in classification included population density (people/km2); ratios of elderly people, agricultural
workers, and people with different educational levels; and the number of physicians per 100,000 people. The urbanization levels were divided into 2 categories on the basis of an NHRI report (levels 1 and 2 represented cities, and levels 3–7 represented rural areas). The monthly costs of patients for insurance premiums were classified into 3 groups, <NT$15,000, NT$15,000 to NT$19,999, and _NT$20,000 (US$1 is approximately
NT$30). Comorbidities diagnosed before the index date included COPD (ICD-9-CM codes 491, 492, and 496), hyperlipidemia
(ICD-9-CM code 272), diabetes (ICD-9-CM code 250), hypertension (ICD-9-CM codes 401-405), congestive heart failure
(ICD-9-CM code 428), liver cirrhosis (ICD-9-CM code 571), and dementia (ICD-9-CM code 290).
Statistical Analysis
The baseline characteristics and comorbidities of the
cohorts were compared. Chi-squared and Student t tests were performed for categorical and continuous variables, respectively. The incidence density of each outcome disease per 1000
person-years was calculated according to sex, age, and comorbidity status. Cox proportional hazard models were used to
estimate the risk of infection outcomes in the treatment cohort compared with those in the untreated cohort. Baseline characteristic
variables, such as age, sex, comorbidities, urbanization
level, and monthly income, were adjusted for in the multivariate model. Hazard ratios (HRs) and 95% confidence intervals (CIs) were estimated using the Cox model.
We further analyzed the frequency of clinical course in the intensive treatment cohort to assess how the responsiveness to the treatment affected outcome risks. The cumulative incidences of outcome diseases were computed using the
Kaplan–Meier method, and the differences between the cohorts were examined using the log-rank test. All statistical analyses were performed using SAS Version 9.3 (SAS Institute, Inc., Carey, NC). Results with a 2-tailed P value of <0.05 were considered statistically significant.
RESULTS
A total of 103,173 ESRD patients receiving HD were
included in our catastrophic illness group (Figure 1). Among them, 42,024 HD patients who had prior diagnosis of PD before HD enrollment were excluded. Patients undergoing HD were categorized into: with PD and received intensive periodontal therapy (n¼6577); and without PD diagnosis (n¼54,242)
between January 1, 1998 and December 31, 2010. After excluding patients on the basis of the aforementioned criteria, the
treatment and comparison cohorts were selected and matched at a 1:1 ratio, according to propensity scores, to reduce selection bias and approximate a randomized trial. Finally, each cohort comprised 4451 patients.
Table 1 compares the baseline characteristics of patients undergoing HD and intensive periodontal treatment with those of the propensity score-matched comparison cohort. The mean dialysis duration was 2.47_2.24 years and 2.50_2.25 years for the treatment and comparison cohorts, respectively. The distribution of sex, age, urbanization level, and CCI was similar in both cohorts. Of the baseline comorbidities, hypertension, hyperlipidemia, and diabetes were prevalent in both cohorts.
Notably, the proportions of liver cirrhosis were high, approximately 27.9% and 28.4% in the treatment and comparison
cohorts, respectively. None of the baseline characteristics differed significantly between the cohorts, except stratified
monthly income.
The overall incidence density of hospitalization for infectious diseases was lower in the intensive treated cohort than in the comparison cohort (48.4 vs 62.5 per 1000 person-years, respectively) (Table 2). We used matched Cox proportional hazards model to analyze the influence of PD therapy on infections. The treated patients exhibited a significantly lower risk of overall infectious diseases than did the comparison cohort, with an HR of 0.72 (95% CI¼0.66–0.78,
P<0.001). Among the infectious diseases for which patients were hospitalized, pneumonia, osteomyelitis, and bacteremia were the primary causes in both cohorts. The incidence rates of pneumonia, osteomyelitis, and bacteremia were 38.5, 6.19, and 7.11 events per 1000 person-years, respectively, in the treated cohort. The treated cohort exhibited a significantly lower risk for acute and subacute IE (HR¼0.54, 95% CI¼0.35–0.84, P<0.01), pneumonia (HR¼0.71, 95% CI¼0.65–0.78, P<0.001), and osteomyelitis (HR¼0.77, 95% CI¼0.62– 0.96, P<0.05) than did the comparison cohort. Although the incidence rates of bacteremia, brain abscess, renal and perinephric abscess were lower in the treated cohort, the risks did
not differ significantly between the cohorts.
We further stratified patients according to age, gender, and comorbidities to estimate the risk difference (Table 3). The intensive treated patients in the age 50 to 64 group exhibited the significantly lower age-specific relative risk of IE (HR¼0.42; 95% CI¼0.22–0.80). In addition, the lower risk for IE was observed in female patients (HR¼0.41, 95% CI¼0.23–0.75), and those with comorbidities (HR¼0.54, 95% CI¼0.35–
0.85). Except patients without comorbidity, the risk of pneumonia in patients with PD treatment in whichever stratification
was significantly lower than that of the comparison cohort. The patients aged elder than 65 years exhibited the lower age-specific relative risk of osteomyelitis (HR¼0.53; 95% CI¼0.34–0.81). For men, the incidence of osteomyelitis were 6.07 and 8.28 per 1000 person-years between the 2 cohorts, with a 0.66-fold relative risk of developing osteomyelitis (95%
in the treated patients with comorbidities than in the corresponding comparison cohort patients (95% CI¼0.62–0.96).
Table 4 presents the effects of treatment responsiveness according to frequency of clinic visit for intensive periodontal treatment. Compared to that in the patients without PD, the risk of hospitalization for overall infectious diseases tended to increase in PD patients who underwent therapy in <1 clinic course, and decreased significantly in patients having more than 2 infective treatment visits, with an HR of 0.52 (95%
CI¼0.48–0.58, P<0.001). The risk of hospitalization for
overall infectious diseases decreased as frequency of treatment visit increased (P value for trend was <0.001). The treatment responsiveness of clinic visit for intensive treatment of PD on IE, bacteremia, pneumonia, and osteomyelitis were evident in the treated cohort.
Because periodontal patients may not have always been
treatment during the study period and this may in fact overestimate the effect of treatment, we also used the Cox proportional
hazard model with time-dependent exposure
covariates to estimate the risk for each outcome disease in order to reduce this bias (Table 5).
The cumulative incidences of major infectious diseases were shown as Kaplan–Meier plot for major infections after 10 years of follow-up (Figure 2). Compared with the control
cohort, the treated cohort exhibited a significantly lower cumulative incidence of IE (Figure 2A) (log-rank test P¼0.008), and
pneumonia (Figure 2C) (log-rank test P<0.001).
DISCUSSION
Primary Results
Our study was the first to demonstrate that PD treatment plays a role in the primary prevention of infectious diseases in patients on HD. We found association between PD therapy and major infectious complications in HD patients. HD patients with PD therapy had a 0.72-fold decrease in overall infectious diseases compared with HD patients without PD, after adjusting for available demographic and medical characteristics.
cohorts, with over 90% patients having CCI score _2. This finding may be explained by the bidirectional relationship between chronic kidney disease and PD, for they share common risk factors such as aging and diabetes.17 In Table 2, among
major infectious complications in both cohorts, the highest incidence rates were observed in pneumonia, bacteremia, and
osteomyelitis, respectively. In Table 3, the risks of IE, pneumonia, and osteomyelitis in patients with comorbidities were
simultaneously lower in PD patients with treatment compared with those of the control cohort.
In Table 4, we observed that the risks among overall
infectious diseases, IE, bacteremia, pneumonia, brain abscess, and osteomyelitis were highest in PD patients with frequency of PD therapy <1 compared to that of control and patients with frequency of PD therapy more than 2. Moreover, the risks among overall infectious diseases, IE, bacteremia, pneumonia, brain abscess, and osteomyelitis were significantly lower in patients with frequency of intensive PD therapy more than 2 compared to that of the control group.
The higher frequency of clinic visit for intensive PD
therapy in patients may represent more completed treatment course or more complex PD burden. In either situation, the HRs of infectious diseases in those who received higher treatment frequency (_2) were lowest compared to that in both control (no PD diagnosis) and lower treatment frequency (_1) patients. These findings further underline the association between the effect of intensive PD therapy and infections in HD patients.
Explanation for the Findings
Pneumonia
Mounting evidence suggests that a causal relationship exists
between PD and respiratory infections such as bacterial pneumonia and COPD.7 Case–
control studies have reported an
increased risk of nosocomial and community-acquired pneumonia and respiratory infections, including COPD exacerbation, in patients with periodontal infections. The risks remained even after adjustment for possible confounding factors, including age and smoking habits.18,19
pneumonia-causing oral pathogens.20 Alternatively, the mechanism
may be the colonization of the upper airway with the assistance of periodontal pathogens, which modify the respiratory epithelium, making the airway more susceptible to the colonization process.21
Oral care strategies for medically compromised patients were devised to prevent aspiration pneumonia.22 Oral care was
believed to reduce lower respiratory tract infections in elderly patients.23 Our study demonstrated that PD therapy was associated
with a 29% reduction in the risk of hospitalization for
pneumonia. The potential effectiveness of PD therapy in preventing pneumonia was comparable to that of influenza vaccination.
24
Infective Endocarditis
Bacteremia in patients with PD tends to be sustained25
and is a major cause of IE in elderly patients with valvular heart disease.26 Recommendations emphasize the use of
antimicrobial prophylaxis before dental procedures to prevent bacterial endocarditis in susceptible patients with periodontitis.27
Studies on intervention therapies, such as decolonization to ameliorate infectious complications in patients on dialysis, further support our findings. A recent population-based study reported that regular dental scaling (at least once per year) reduced the risk of IE by 33%,12 probably because scaling can
reduce the bacterial load and cause a shift in the subgingival flora.28 Although our study did not reveal a significant reduction
in the risk of bacteremia in the treated group, the incidence rates were lower than that of the control cohort.
Osteomyelitis
ESRDis a comorbidity that predisposes patients to vertebral osteomyelitis (VO).28 HD has been associated with the highest
rate of in-hospital mortality in patientswithVO.29VOoccurrence
was associated with antecedent infections, DM, and immunosuppression
30; however, few studies have examined the relationship
between periodontal infection and VO.
Although the primary entry sites of the hematogenous
pathogens in osteomyelitis included soft tissue, vascular access sites, and endocarditis; the pathogens were identified in only
51% of patients.31 Our result revealed that the risk of bacteremia
tended to decrease in HD patient with PD treatment, and it might infer that oral cavity being another potential route of entry in VO patients.
Limitations
The strengths of our study are that we used longitudinal,
population-based data to demonstrate the demographic characteristics of patients undergoing HD with PD. Our study is the
first to demonstrate an association between intensive PD intervention and major infectious diseases in high-risk patients with
HD. These findings can be generalized to the overall ESRD population and may be aid in alerting clinicians or policy makers to the role of intensive PD therapy in the primary prevention of infectious diseases.
This observational study was performed using administrative databases and has inherent limitations.
First, our study was not a prospective, randomized study of hemodialysis (HD) patients with PD who did or did not receive adequate periodontal care. Although inherent selection bias could not be avoided, the significant risk reductions in the
patients with higher frequency of intensive PD treatment compared with that of the patients without PD further confirm the
role of PD therapy in the primary prevention of infectious diseases. Nevertheless, future studies involving well-designed, controlled interventions are required.
Second, we are uncertain as to the PD status of the control cohort. In HD patient without PD diagnosis coding, the reasons might be: patients did not have PD; or patients were underdiagnosed for PD. According to a prospective observational incenter
study conducted by Chen et al, 80.6% of prevalent HD
patients have PD,1 which further confirms our control cohort as
a mixed population of whom up to 80% had undiagnosed (and untreated) peridontal disease. Oral disease may be underdiagnosed in patients treated with dialysis due to the lower uptake of
public dental service.32,33 As we can see in our study, the CCI
_3 were approximately 53% in both cohorts (Table 1). The high prevalence of comorbidities might prevent patients from dental care. This is the reason we chose patients without PD diagnosis
(n¼54,242) in our database as control cohort. Finally, the NHIRD does not contain detailed smoking
habit, weight, or other lifestyle factors that could be potential confounders. However, recent evidence has indicated that the observed association between PD and atherosclerotic vascular disease may exist independently of smoking.34 We adopted
other important variables such as socioeconomic status and access to dental facilities (urbanization level) to strengthen the validity of our result.
CONCLUSION
Our study confirm that intensive periodontal treatment is associated with a significant risk reduction in major infectious complications in HD patient with PD. PD therapy is a potential modifiable factor in the primary prevention of infections in HD patients, especially in patients with multiple comorbidities. Therefore, the diagnosis and management of PD in HD population deserve better awareness.