Associations between C-reactive protein and all-cause mortality among oldest old adults in Chinese longevity areas: A community-based cohort study

Background: Higher C-reactive protein (CRP) levels have been proposed as a predictor of all-cause mortality in many existing studies from multiple populations, but the association for the oldest old adults (aged 80 and older) remains unclear. Objective: To examine the association between CRP and all-cause mortality among the oldest old Chinese adults. Design: This is a prospective, community-based longitudinal cohort study with 2206 adults aged 80 years old and older with available CRP test results. Cox proportional hazards regression models were used to estimate hazard ratios (HRs) with 95% condential intervals (95% CIs) for all-cause mortality according to CRP quartiles, adjusting for potential confounders. Results: The median age of the participants was 93 years old, and the median CRP concentration was 1.13 mg/L at baseline. During a median follow-up period of 36.7 months, 1106 deaths were veried. After full adjustment for potential confounders, a high CRP concentration was positively associated with an increased risk of all-cause mortality. Compared with the lowest quartile, the fully adjusted HRs of the second, third, and fourth quartiles were 1.17 (95% CI: 0.94, 1.46), 1.28 (95% CI: 1.01, 1.61), and 1.50 (95% CI: 1.20, 1.87), respectively. The association of CRP with all-cause mortality was likely modied by smoking status (P = 0.011). Conclusions: Our study indicated that a high CRP concentration was likely to be a prospective factor predicting death among the oldest old adults. Future studies investigating additional factors of disease and aging processes are needed to obtain a better understanding of the mechanisms. This on the datasets from the Chinese Longitudinal Longevity Survey (CLHLS) The CLHLS datasets are available at the Computerized Researchers can obtain these data after submitting a data use agreement to the CLHLS team.


Background
Inflammation has been studied to be the role of a wide range of aging-related diseases [1], such as atherosclerosis and coronary artery disease [2], diabetes [3], Alzheimer's disease [4] and cancer [5]. In ammaging, a description of low-grade, chronic, systemic in ammation in aging, is a highly signi cant risk factor for both morbidity and mortality in elderly people, as most if not all age-related diseases share in ammatory pathogenesis. Nevertheless, the precise etiology of in ammaging and its potential causal role in contributing to adverse health outcomes remain largely unknown [6,7]. Chronic, low-grade elevations in markers of in ammation, such as C-reactive protein (CRP), are potent risk factors for allcause mortality [8].
CRP, an acute-phase protein produced predominantly by hepatocytes, is a sensitive and exquisitely systemic marker of inflammation [9]. CRP has been commonly assayed for infections [10], in-hospital complications [11], prognosis in uences [12] and aging-related health outcomes in clinical applications, especially cardiovascular and metabolic disease risk [13,14].Higher CRP levels have been proposed as a predictor of all-cause mortality in many [15][16][17][18][19][20][21][22][23][24][25][26] but not all studies [27]. Inconsistent results may exist due to sex, ethnic or age differences in the populations, and the strength of the association also varied across studies. Moreover, these ndings are based on the general population, but the oldest old adults (octogenarians, nonagenarians and centenarians) remain underrepresented. The classic risk markers for disease and mortality might not be effective in the oldest old population [28]. Therefore, we conducted the present study to prospectively examine whether CRP is associated with allcause mortality among the oldest old adults based on datasets from the Chinese Longitudinal Healthy Longevity Survey (CLHLS) in longevity areas.

Methods
Design, study setting, and participants This is a perspective, community-based longitudinal cohort study. Participants were recruited in the sixth wave (2012) and the seventh wave (2014) of CLHLS from eight longevity areas selected by the Chinese Society of Gerontology. The densities of oldest old adults are higher (especially for centenarians) in longevity areas than in other areas. These areas include Chen Mai County (Hainan Province), Yong Fu County (Guangxi Province), Ma Yang County (Hunan Province), Zhong Xiang City (Hubei Province), Xia Yi County (He Nan Province), San Shui City (Guangdong Province), Lai Zhou City (Shandong Province), and Ru Dong County (Jiangsu Province). Overall, 2206 participants were enrolled at baseline and follow-up in 2014 and 2017, respectively. We included all adults aged 80 years or older with available results of CRP tests, and 269 of these adults were lost to follow-up ( Figure 1). The study was approved by the biomedical ethics committee of Peking University. All participants included in CLHLS provided informed consent. More details of CLHLS have been previously described [29].

Measurement of CRP
Venous blood samples were obtained from the participants by collecting in heparin anticoagulant vacuum tubes. CRP concentration was generally measured through a high-sensitivity immunoturbidimetry assay, and all blood biochemistry tests were conducted by the central clinical lab at Capital Medical University in Beijing.

Measurement of all-cause mortality
We veri ed the survival status of all participants at baseline during follow-up surveys in 2014 and 2017.
Date of death was inquired and ascertained from family members or caregivers of the deceased. The survival time for participants was calculated from the date they enrolled in our study to the date of death.
For survival, survival time was identi ed as right-censored at the date of the latest follow-up. Those who could not be found and contacted were recorded as "lost to follow-up".

Measurement of covariates
Covariate information was collected via face-to-face structured questionnaires and biochemistry assays.
Covariates in our analyses included sociodemographic information (age, sex, education and residence), Dietary habits include vegetable intake, fruit intake, meat intake, and exercise. For the frequencies of food intakes, "almost every day" or "often" were categorized into "often" and "occasionally" and "rarely or never" was categorized into "not often"; for exercise, "yes" or "no" was determined from the question, "Do you do exercises regularly at present?". MMSE [30] is a practical scale for grading the cognitive state, and the oldest old adults in China with MMSE scores below 24 could be de ned as having cognitive impairment [31]. Frailty status was classi ed according to the Study of Osteoporotic Fractures (SOF) index [32] using three components as follows: 1) weight loss (BMI < 18.5 kg/m 2 ); 2) inability to rise from a chair without using arms; 3) reduced energy level, de ned by a "yes" response to the question, "For at least the last 6 months have you been limited in activities people usually do, because of a health problem?" The status was categorized as robust (no components), prefrail (1 component) or frail (2 or 3 components), which has been shown to be an applicable indicator of biological age in Chinese older adults [33]. For the medical history, hypertension was de ned as SBP≥140 mmHg and/or DBP≥90mmHg based on 2018 Chinese guidelines for the management of hypertension [34]; diabetes was de ned as fasting blood glucose≥7.0 mmol/L based on National guidelines for the prevention and control of diabetes in primary care [35]for the Chinese population; CVD was determined by the self-report of the participants.

Statistical analysis
A table for baseline characteristics was generated using descriptive statistics strati ed by CRP quartiles (mg/L). Continuous data were described by medians and interquartile ranges (IQR), and categorical data were described by frequencies and percentages (%). Hypotheses regarding differences in characteristics across quartiles of CRP were analyzed using linear regression for continuous variables and χ 2 tests for categorical variables.
Kaplan-Meier curves were generated for the quartiles of CRP concentrations, and log-rank tests were used to compare different quartile subgroups. Cox proportional hazards regression models were used to estimate hazard ratios (HRs) with 95% con dential intervals (95% CIs) of mortality by CRP quartiles, with the lowest quartile (Q1) as the reference group. The Cox models were adjusted for potential confounders that may be associated with both CRP concentrations and mortality. The following three models with different adjustments were used: 1) the rst model (model 1) tested the association between CRP and mortality, controlling for age and sex; 2) the second model (model 2) was further adjusted for other baseline characteristics, namely, education time (0 year or ≥1 year), residence status (rural or urban), smoking status (current or not current), alcohol consumption (current or not current), vegetable intake (often or not often), fruit intake (often or not often), meat intake (often or not often), and exercise (yes or no); and 3) the third adjusted model (model 3) was further adjusted for physical examination, disease status and biochemical indicators. The aforementioned covariates included BMI (continuous), MMSE scores (continuous), frailty status (frail, prefrail, robust), hypertension (yes or no), diabetes (yes or no), CVD (yes or no), cholesterol (continuous), and triglycerides (continuous). Model 3 was considered to be fully adjusted. Tests of linear trends were performed by treating the median values for each quartile of CRP as a continuous variable.
Analyses were conducted using Stata version 14.0 (College Station, Texas). A P-value<0.05 was considered statistically signi cant.

Baseline characteristics
Among 2206 individuals, the median age of participants was 93 years (IQR: 86-100 years). A total of 1417 were women (64.23%), and 1905 were living in rural areas (87.19%). Baseline characteristics are summarized in Table 1 Table 2 presents the association between CRP and mortality. Compared with the lowest quartile, the fully adjusted HRs of the second, third, and fourth quartiles were 1.17 (95% CI: 0.94, 1.46), 1.28 (95% CI: 1.01, 1.61), and 1.50 (95% CI: 1.20, 1.87), respectively ( Table 2). The risk of all-cause mortality increased with elevated CRP (P < 0.001). A supplementary analysis was also conducted based on the recommendation for relative risk categories of CRP levels

Subgroup analyses
Subgroup analyses strati ed by major confounders are presented in Table 3. The HRs showed similar results with no signi cant differences across most subgroups de ned by age, sex, education, residence, drinking status, vegetable intake, fruit intake, meat intake, exercise, BMI, frailty status, MMSE scores. However, a signi cant interaction from smoking status was noted (P = 0.011).

Discussion
In this population-based 6-year follow-up study of oldest old adults living in Chinese longevity areas, the participants with high CRP concentrations had an increased risk of mortality, even after adjusting for potential confounders. Strati ed by quartiles, the highest quartile of CRP showed a statistically signi cant difference compared with the lowest quartile after full adjustment. The association of CRP with all-cause mortality was less likely modi ed by sociodemographic factors, physical examinations, biochemical indicators and most lifestyle factors, except for smoking status.
Our ndings are consistent with previous studies demonstrating positive associations between CRP and all-cause mortality, which are signi cant at higher levels of the CRP distribution [15,[19][20][21][23][24][25][26]. The estimated value might differ by ethnicity because Asian populations tend to have a lower CRP level than Western populations [37,38]. A study [39] with 11623 middle-aged Chinese individuals categorized three groups based on CRP levels (< 1.0, 1.0-3.0, and > 3.0) and obtained the result that the HR for all-cause mortality in the > 3.0 group was 2.64 (95% CI: 1.74, 4.01). This difference inferred that the estimate might wane with age or that the sample size of our current study was not su cient to provide power to detect a difference.
It is also important to note that sex was not a modi er in our study. Some studies showed a positive association in both sexes [8,23], while signi cant differences appeared to exist in a single sex (mostly male [27,40,41]). However, whether males or females are at a greater risk remains controversial [8,23,41,42]. The differential effect of CRP in predicting all-cause mortality risk by sex warrants further investigation. Smoking status is another novel point in which signi cant interaction was found. However, the estimate seemed to be stronger in nonsmokers, though both strati cations showed no signi cant differences. One explanation is that in ammation adaption might occur in the human body during the period of habitual smoking, resulting in a lower hazard to current smokers than to those who did not smoke during the same period. From another consideration, a limited sample size leads to a lack of statistical signi cance in the estimate, and even interaction exists. Moreover, evidence shows that smoking cessation does not reduce CRP [43], but in our study, we did not take this into account during grouping, which might cause misclassi cation bias. The interaction observed by smoking status warrants further research.
Potential limitations of the current study should be considered in evaluating our results. Our study is observational in nature, and we cannot rule out the possibility of reverse causality; therefore, CRP might also be a consequence of diseases rather than a cause. Moreover, the residual confounded by other unmeasured or unknown factors likely exists and potentially results are biased in an unknown direction despite our full adjustment in analyses. Additionally, since information on the subtype of death was not collected in the CLHLS, in-depth analyses based on cause-speci c mortality are necessary but unable to conduct. Finally, similar to most other studies, the fact that CRP was measured only once at baseline is a potential limitation because random fluctuation in this parameter over time would tend to increase the variance in the data; how trajectories of CRP may in uence mortality remains undetermined.
Despite these limitations, this study has noteworthy strengths when compared to prior research. Above all, our ndings were based on a prospective study with integrated and detailed baseline, outcome, and blood sample data. The robustness of the outcomes measured and the large sample size of the oldest old adults increases the relevance of our ndings. For representativeness, it is believed that communitydwelling older adults are more typical due to the dominance of family care in the Chinese society. A distinguishing feature of this study is that all of the longevity areas we investigated provided a distinct population of oldest old adults, which broadens the evidence from existing research with a unique age spectrum.

Conclusion
Our analyses indicated that high CRP concentrations are likely to be a prospective factor predicting death among the oldest old adults. CRP may be more useful clinically in identifying higher risk populations for all-cause mortality. Future studies investigating additional factors of disease and aging processes are needed to conduct a better understanding of the mechanisms. Abbreviations BMI, body mass index; CI, con dential interval; CLHLS, the Chinese Longitudinal Healthy Longevity Survey; CRP, C-reactive protein; CVD, cardiovascular disease; DBP, diastolic blood pressure; HR, hazard ratio; IQR, interquartile range; MMSE, Mini-Mental State Examination; SBP, systolic blood pressure; SOF, Study of Osteoporotic Fractures.

Ethics approval and consent to participate
The biomedical ethics committee of Peking University approved the study (IRB00001052-13074). All participants signed written informed consent.

Availability of data and material
This study was based on the datasets from the Chinese Longitudinal Healthy Longevity Survey (CLHLS) in longevity areas. The CLHLS datasets are publicly available at the National Archive of Computerized Data on Aging (NACDA), University of Michigan (https://www.icpsr.umich.edu/icpsrweb/NACDA/series/487). Researchers can obtain these data after submitting a data use agreement to the CLHLS team.

Competing interests
The authors declare that they have no competing interests.

Funding
The Chinese Longitudinal Healthy Longevity Study (CLHLS), which provided the data analyzed in this paper, is jointly supported by the National Natural Sciences Foundation of China (81573207, 71233001, 71490732, and 81573247), and the U.S. National Institute of Aging (2P01AG031719 and 3P01AG031719-07S1). This work was also supported by the National Key Research and Development Program of China (2018YFC2000400), the Construction of High-level University of Guangdong (C1050008 and C1051007), and the National Institutes of Health (NIH/NIA P30-AG028716). The funders played no role in study design or implementation; data collection, management, analysis, and interpretation; manuscript preparation, review, or approval; or the decision to submit the manuscript for publication.
Authors' contributions XBW, XMS and CM designed the study analysis. ZJC, ZF, YBL, YLQ, LL, YCL and JHZ conducted CLHLS and directed its implementation, including quality assurance and control, dataset management and analytic strategy. ZXR and FRL contributed to data cleaning. YBL and ZXY helped supervise the eld activities and designed the study's analytic strategy. PLC, HLY, and XC analyzed the data and prepared the manuscript. All authors have critically commented on and revised the manuscript, and approved the nal version. XBW and XMS are guarantors of the paper.   Kaplan-Meier graphs for all-cause mortality by quartiles of CRP