Moderately Reduced Estimated Glomerular Filtration Rate (eGFR) among Community - Dwelling Older People Associated with Increased Long Term Mortality: A Study from Low to Middle Income Country in Asia.

Background: Reduced glomerular ltration rate (GFR) is regarded as a component of renal senescence; however, the full impact of chronic kidney disease (CKD) among older people is still not fully understood. This study aimed to investigate the effect of reduced GFR on all-cause mortality, and the combined effect of socioeconomic status and CKD on mortality among older Thai population. Methods: A retrospective cohort of Thai population aged >60 years was enrolled from the Fourth National Health Examination Survey, and estimated GFR (eGFR) and comorbid disease data were collected. Mortality information was gathered from the National Civil Registration and Vital Statistics System. Factors associated with all-cause mortality were determined using Cox proportional hazards model. Results: A total of 9,080 participants (mean age: 69.5 years) were included. Prevalence of CKD stage 3-5 was 53.3%. During the 2008-2015 period, the overall mortality rate among CKD was 28.9 per 1,000 person-years. The stages of CKD signicantly associated with increased mortality were stages 3b, 4, and 5, as follows: HR: 1.4 (95% CI: 1.2-1.63), 2.2 (95% CI: 1.78-2.71), and HR: 5.9 (95% CI: 4.26-8.15), respectively. People with higher socioeconomic status and moderate-stage CKD had lower mortality than lower socioeconomic status moderate-stage CKD. Conclusion: Moderate decline in eGFR was found to be independently associated with all-cause mortality in Thai elderly population. Socioeconomic disparities were also observed among older CKD. Implementation of a mass screening program to identify asymptomatic individuals would accelerate diagnosis and treatment, and this would reduce the costs of care associated with advanced-stage disease.

in estimated GFR (eGFR). Moreover, early-stage CKD was shown to associate with lower mortality in older adults compared to younger adults with CKD. (10) It has been suggested that the majority of older patients with CKD did not transition to ESRD, but that they died from other causes. Although the association between socioeconomic status and the prevalence of CKD is well established (11) , the socioeconomic disparity in CKD with regard to mortality risk was demonstrated only in high income countries (11,12) .
Evidence speci c to the clinical impact of CKD stagng on mortality among older people in LMICs remains comparatively scarce.
Thailand is a middle-income country with an increasing prevalence of cardiometabolic diseases. This trend during the last decade is supported by data from recent national health examination surveys (13)(14)(15) . Increases in related risk factors and the average age of the population suggest an increasing prevalence of CKD in Thailand. Previous studies from Thailand (16,17) reported a prevalence of CKD among older people of approximately 40%, which is substantially higher than the rate of CKD among younger population. However, those studies did not explore the impact of CKD on mortality among older population, and whether any identi ed association is in uenced by socioeconomic status. Accordingly, the aim of this study was to investigate the effect of reduced GFR on all-cause mortality, and the combined effect of socioeconomic status and CKD on mortality among older persons in Thailand.

Methods
A retrospective cohort of elderly Thai population aged 60 years or more from the Fourth Thai National Health Examination Survey (NHES-IV) conducted during 2008-2009 were evaluated in 2016 to determine all-cause mortality. NHES-IV was a nationally representative survey using a multistage, strati ed sampling of the Thai population for which the detailed of sampling methods have been described (14) .
Brie y, the rst stage of sampling units was provinces in each region in Thailand, the second stage was district in each province, the third was electoral unit (EUs) or villages, and, the last stage was age and gender-speci c individuals. Medical comorbid disease (including eGFR), functional status, and socioeconomic status data were collected at the baseline assessment in NHES-IV. The vital status of each participant was ascertained from the National Civil Registration and Vital Statistics System.
Diabetes was de ned as fasting plasma glucose of >7.0 mmol/L, diagnosis by a physician, or taking antidiabetic medications. (20) Hypertension was de ned as systolic blood pressure >140 mmHg, diastolic blood pressure >90 mmHg, or history of using any antihypertensive medications. Hypercholesterolemia was de ned as total serum cholesterol >5.2 mmol/L or use of cholesterol-lowering medication within the previous 2 weeks. According to WHO criteria, anemia was de ned as hemoglobin level <13 g/dl in males, and <12 gm/dl in females (21) .
Data collected in the NHES-IV, including demographic, socioeconomic, and health data such as smoking history, medication used, and common medical conditions, using semi-structured interview with standardized questionnaires. Medical comorbidities inquired in the interview included history of diabetes, hypertension, high cholesterol, asthma, chronic airway disease, allergy and cerebrovascular disease. Data relating to activities of daily living (ADLs) for both basic ADLs (B-ADLs) and instrumental ADLs (I-ADLs) were also obtained by interviewing enrolled participants and relatives. Impairment in ADLs was de ned as the requirement for any assistance in performing ADL-related tasks, which included feeding, dressing, bathing, toileting and transferring from beds or chairs. I-ADL impairment was de ned as the requirement for any assistance in complex tasks including using telephone, handling nanced, responsibility for own medication, using transportation or driving and housework. The wealth index score was determined by assessing the number of household items owned, and then a standardized score was assigned to each participant (22) . Socioeconomic status (SES) was strati ed into 5 quintiles, with the lowest quintile indicating the poorest group, and the highest quintile indicating the wealthiest group.
Mortality data was retrieved until May 2016 from the Thai Vital Registration System, Bureau of Registration Administration, Ministry of Interior, which was a rmed to be a reliable source of death statistic in Thailand. (23) Statistical analysis Descriptive statistics were used to compare baseline characteristics of participants. Categorical data are reported as frequency and proportion, and continuous data are given as mean ± standard deviation or median with minimum and maximum depending on the distribution of data. Analysis of variance (ANOVA) test was applied for continuous variables, while chi-square test was used for categorical variables. A p-value less than 0.05 was considered statistically signi cant. In order to consider for the multistage, strati ed sampling survey design of the NHES-IV, the analysis was carried out by using svyset syntax to adjusted for all stage of sampling and nite population correction. The adjusted data was saved for further analysis. Cox regressions were conducted to determine the unadjusted hazard ratios (HRs) for mortality. Potential confounders of all-cause mortality among participants with CKD included in the analysis were age, gender, current smoking status, regular analgesic use, hypertension, diabetes, hypercholesterolemia, history of stroke, anemia, and wealth index score. Variables that had a p <0.1 in univariate analyses were then included in a Cox proportional hazards model. Statistical analyses were performed using STATA 15.0 (StataCorp LP, College Station, TX, USA).

Results
Patient demographic data and prevalence of CKD A total of 9,080 participants with a mean age of 69.5±7.0 years were included. The mean age of patients with CKD stage 3-5 was higher than that of people without CKD. The prevalence of CKD stage 3-5 in this cohort was 53.3%, and the prevalence of CKD by stage was 34.0%, 14.5%, 4.0%, and 0.8% for stage 3a, 3b, 4, and 5, respectively. The prevalence of CKD stage 3-5 increased with age, as follows: 44.2% for age group 60-69, 61.2% for age group 70-79, and 76.3% for age group 80 years or more. Comorbid diseases including diabetes, hypertension, hypercholesterolemia, anemia and stroke, were reported at an increasingly higher proportion as the stages of CKD becoming more advanced. Baseline characteristics of participants by CKD stage based on GFR category are shown in Table 1.
Estimated GFR decreased with increasing age, particularly among participants with diabetes mellitus and/or hypertension. Further analysis was conducted among participants aged 85 years or older without diabetes and hypertension, and the mean eGFR was 53.4 ml/min/1.73m 2 , which corresponds with CKD stage 3 according to the current guideline.

Mortality and CKD staging
During a total follow-up time of 52,120.8 person-years, 1,505 (15.3%) participants died for an overall mortality rate of 28.9 per 1,000 person-years (95% con dence interval [CI]: 27.5-30.4). Among CKD subjects, the mortality rate increased with more advanced stage of CKD from 27.2 per 1,000 person-years (95% CI: 24.9-29.7) in CKD stage 3a to 167.3 per 1,000 person-years (95% CI: 124.1-225.6) in CKD stage 5 (Table 2). Survival curves compared among the different stages of CKD are illustrated in Figure 1.

Factors associated with increased mortality
The factors associated with all-cause mortality from univariate and multivariable analyses are shown in Table 3. In univariate analysis, age, male, hypertension, diabetes, hypercholesterolemia, stroke, anemia, smoking, regular used analgesic drugs and all stages of CKD were signi cantly associated with increased mortality risk. Better socioeconomic status was shown to be a protective factor against mortality. Subsequent multivariate analysis revealed age, male, diabetes, hypercholesterolemia, stroke, anemia, smoking and regular used analgesic drugs to be independent risk factors for mortality. CKD stages 3b to stages 5 were also independently associated with increased risk of mortality in a dose-response manner, as follows: hazard ratio [HR]: HR: 1.4 (95% CI: 1.2-1.63) for stage 3b; HR: 2.2 (95% CI: 1.78-2.71) for stage 4; and, HR: 5.9 (95% CI: 4.26-8.15) for CKD stage 5. Our analysis also showed socioeconomic status to be independently associated with mortality. Speci cally, the higher the wealth index, the lower the person's risk of death.

Discussion
The present study demonstrates CKD to be an important risk factor for increased mortality among older people, and the strongest association with mortality is among patients with advanced-stage CKD. Three studies have explored the effect of CKD on mortality among community-dwelling older people (24)(25)(26) , and all of those studies were conducted in high income countries. CKD was identi ed as an independent risk factor for mortality in two of those studies (24,26) , but not in the one other study (25) . The present study con rms that CKD plays an important role in predicting mortality among older people in a middle-income country, and that relationship was shown to be statistically signi cant starting at CKD stage 3b and higher. The observed associations were in a dose-response manner, with stronger associations observed as the stage of CKD became higher. The observed association between CKD and mortality among the elderly is consistent with recently reported ndings that older people with CKD showed higher metabolic complications, cognitive impairment, frailty, and all-cause mortality. (27,28) A key reason for changing the de nition of CKD stemmed from reported evidence suggesting association between reduced eGFR and increased risk of death (7,29) . However, when we speci cally focused on older population at an eGFR level of 60 ml/min/1.73m 2 , which is consistent with CKD stage 3a, the association with mortality was fairly weak in younger older people, and non-signi cant in people aged older than 75 years. Nevertheless, this association become more apparent at lower levels of eGFR (29) . The present study also found a similar trend that lower eGFR, starting at CKD stage 3b, is associated with increased mortality. Our nding is also consistent with that from a previous study (30) that found CKD stage 3b to be signi cantly associated with increased mortality. This nding has important clinical implications for managing older people with multiple chronic comorbidities. Inaccurate diagnosis or staging of CKD would result in extra psychological concern for the patient, and the burden and costs of repeated assessment and potentially unnecessary treatment (7) . However, failure to identify and treat a disease would likely lead to increased morbidity and premature death. Our nding that CKD stage 3b-5 is clinically signi cant in older adults should heighten our attention to the need for an improved pathway of care for these patients.
Regarding the relationship between socioeconomic status and mortality among older people with CKD, the present study found the socioeconomic status of people in the 2 highest wealth quintiles to be signi cantly correlated with lower all-cause mortality among participants with CKD stage 3b or higher in both univariate and multivariate analysis. Lower socioeconomic status is well recognized for its association with higher prevalence of unfavorable outcomes of CKD in high income countries (11,12) .
Evidence speci c to the effect of disparities in socioeconomic status on CKD patients in low to middle income countries is limited, particularly relative to the association between socioeconomic status and clinical outcomes (11) . The present study demonstrated association between lower socioeconomic status and clinical disadvantages among patients with CKD stage 3b-5 in a middle-income country setting, which is similar to the ndings in high income countries. Association between socioeconomic status and CKD might be due to the fact that non-communicable diseases (NCDs), chronic in ammation, neurohormonal activation, and oxidative stress are all more prevalently observed among people with a lower socioeconomic status (12) . Moreover, lower socioeconomic status may also be associated with limited access to healthcare services and substandard living conditions that contribute to the development of CKD, disease progression, and premature mortality (12) .

Limitations
This study has some mentionable limitations. First, the fact that CKD was de ned via one-time serum creatinine testing without urine albumin level testing could cast some doubt over the true prevalence of CKD in our study population. However, the fact that CKD stage 3b, stage 4, and stage 5 were found to be strongly associated with increased mortality in multivariate analysis demonstrates the validity and clinical importance of the method used in our study. Second and last, we did not evaluate data speci c to the progression of CKD to ESRD. This evolution is complicated by the fact that older people may die from other causes before developing and/or succumbing to ESRD. Understanding this course after adjusting for related confounders would further improve our understanding of the relationship between CKD and mortality among older population. The strengths of this study include the large size of the study population and the fact that sampling was nationwide. Moreover, representative sampling of communitydwelling older persons was employed to estimate the prevalence of the condition. Our study also explored the association between CKD and all-cause mortality in a low to middle income country. Although a number of studies have investigated the prevalence of CKD, no studies have evaluated the impact of CKD on clinical outcomes in a low to middle income country (1) .

Conclusion
Moderate decline in eGFR was found to be independently associated with all-cause mortality in community-dwelling elderly Thai population. Socioeconomic disparities were also observed among older CKD, with higher mortality observed among poorer people. Implementation of a mass screening program to identify asymptomatic individuals would accelerate diagnosis and treatment, and this would reduce the costs of care associated with advanced-stage disease. The study was approved by Institutional Review Board, Siriraj hospital, Mahidol University. All participants had signed the informed consents before included in this study.

Consent for publication
Not applicable.

Availability of data and materials
The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Competing interests
The authors declare no competing interests.

Funding
No funding was obtained for this study.