This study utilized data from the CHARLS, a nationally representative longitudinal survey of Chinese adults aged 45 years and older. CHARLS was initiated in 2011 and includes biennial follow-ups to collect detailed information on demographic, socioeconomic, lifestyle, and health-related factors [23]. For this analysis, we included participants who completed the baseline survey in 2011 and were followed up until 2020. For the two separate analyses, participants with prevalent CKD at baseline were excluded from the analysis assessing the risk of incident CKD associated with baseline CLD, and participants with prevalent CLD at baseline were excluded from the analysis assessing the risk of incident CLD associated with baseline CKD. The final sample sizes were 3651 and 5530 for the two analyses, respectively (Fig. 1).

Fig. 1.

Participant screening flowchart. a) CLD increases incident CKD risk; b) CKD increases CLD risk.

*Note: We also utilized the 3rd wave of cystatin C to calculate eGFR to synthesize CKD onset and therefore excluded dissimilarity in numbers here.

CLD was defined based on self-reported physician diagnoses of chronic liver conditions, including hepatitis, cirrhosis, or other chronic liver diseases. CKD was defined using a combination of self-reported physician diagnoses and laboratory data. Estimated glomerular filtration rate (eGFRcysc) was calculated using the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equation by cystatin C [24]. CKD was defined as eGFRcysc < 60 ml/min/1.73 m², consistent with international guidelines [25]. Incident cases of CLD and CKD were identified during follow-up based on new self-reported diagnoses or laboratory measurements meeting the above criteria.

A wide range of demographic, lifestyle, and clinical factors were included as covariates in the analysis. This included age, sex, education level (categorized as illiterate, primary school, middle school, or higher), and marital status (married or not). Lifestyle factors included sleep duration (hours per night), body mass index (BMI, calculated as weight in kilograms divided by height in meters squared), smoking status (current smoker, former smoker, or never smoker), and alcohol consumption (current drinker, former drinker, or never drinker). Clinical factors included the presence of hypertension (self-reported diagnosis or systolic blood pressure ≥ 140 mmHg or diastolic blood pressure ≥ 90 mmHg) [26], diabetes (self-reported diagnosis or fasting blood glucose ≥ 7.0 mmol/l) [27], hyperlipidemia (self-reported diagnosis or elevated total cholesterol ≥ 5.2 mmol/l) [28], and heart disease (self-reported diagnosis of coronary heart disease, myocardial infarction, or other heart conditions) [29], and baseline eGFRcysc.

Baseline characteristics of the study population were summarized using median and interquartile for continuous variables and frequencies and percentages for categorical variables. Differences between groups were assessed using t-tests or chi-square tests as appropriate. Cox proportional hazards models were used to estimate hazard ratios (HRs) and 95 % confidence intervals (95 % CIs) for the association between baseline CLD and incident CKD, as well as between baseline CKD and incident CLD. Time to event was defined as the time from baseline to the first occurrence of the outcome, loss to follow-up, or the end of the study period in 2020, whichever came first. The proportional hazards assumption was tested using Schoenfeld residuals.

Models were adjusted for potential confounders, including age, sex, education, marital status, sleep duration, BMI, smoking, alcohol consumption, hypertension, diabetes, hyperlipidemia, heart disease, and baseline eGFRcysc. Missing data were handled using random forest imputation techniques to minimize bias and improve robustness. Subgroup analyses were performed to examine potential effect modification by key covariates, such as diabetes and hypertension. To assess latent effects, we conducted sensitivity analyses by excluding the first 2 years of follow-up.

All statistical analyses were conducted using Stata version 16.0 (StataCorp, College Station, TX) and R software (version 4.2.0). A two-sided P-value < 0.05 was considered statistically significant.

The study protocols were approved by the Ethical Review Committee of Peking University (CHARLS: IRB00001052–11015).

The baseline characteristics of included population are summarized in Table 1. For the analysis assessing the association between CLD and incident CKD, a total of 3651 participants were included, of whom 123 (3.4 %) had CLD at baseline. For the analysis assessing the association between CKD and incident CLD, 5530 participants were included, with 1092 (19.7 %) having CKD at baseline. Participants with baseline CLD were more likely to be male (47.2 % vs. 42.4 %), current smokers (29.3 % vs. 27.8 %), and past drinkers (15.4 % vs. 7.1 %) compared to those without CLD. They also had a higher prevalence of diabetes (12.2 % vs. 5.9 %) and hyperlipidemia (21.1 % vs. 10.2 %). Similarly, participants with baseline CKD were older (median age: 66 vs. 57 years), more likely to be female (52.3 % vs. 42.8 %), and had a higher prevalence of hypertension (32.2 % vs. 24.0 %), diabetes (6.3 % vs. 5.3 %), and heart disease (15.8 % vs. 10.4 %) compared to those without CKD. Baseline eGFRcysc was significantly lower in participants with CKD (median: 55 vs. 84 ml/min/1.73 m², P < 0.001).

Among participants without CKD at baseline, a total of 575 incident cases of CKD occurred during a median follow-up of 9.0 years, with an overall incidence rate of 18.68 per 1000 population (95 % CI: 17.16–20.21) (Table 2). Stratified by CLD status, participants with CLD exhibited a significantly higher incidence rate of CKD (37.25 per 1000 population, 95 % CI: 25.08–49.42) compared to those without CLD (18.08 per 1000 population, 95 % CI: 16.55–19.61). In crude model, CLD was associated with an increased risk of CKD (HR=2.09, 95 % CI: 1.49–2.93, P < 0.001). This association remained consistent after adjustment for age and sex (HR=2.09, 95 % CI: 1.49–2.93, P < 0.001) and further adjustment for multiple confounders in Model 3 (HR=1.93, 95 % CI: 1.37–2.72, P < 0.001) (Table 2).

Conversely, among participants without CLD at baseline, 474 incident cases of CLD were identified, with an incidence rate of 9.80 per 1000 population (95 % CI: 8.92–10.68) (Table 2). Participants with CKD at baseline experienced a significantly higher incidence rate of CLD (13.56 per 1000 population, 95 % CI: 11.21–15.90) compared to those without CKD (8.89 per 1000 population, 95 % CI: 7.95–9.83). CKD was independently associated with an increased risk of developing CLD in all models (Model 1: HR=1.56, 95 % CI: 1.27–1.91, P < 0.001; Model 2: HR=1.72, 95 % CI: 1.38–2.14, P < 0.001; Model 3: HR=1.68, 95 % CI: 1.31–2.16, P < 0.001) (Table 2). Excluding CLD cases from the first two years of follow-up did not change these findings (Table 3).

The Kaplan-Meier survival curves illustrate the cumulative incidence of CKD stratified by CLD status and the cumulative incidence of CLD stratified by CKD status. Participants with baseline CLD showed a significantly higher risk of CKD development compared to those without CLD (log-rank P < 0.001). Similarly, participants with baseline CKD demonstrated a significantly higher risk of CLD occurrence compared to those without CKD (log-rank P < 0.001) (Fig. 2).

Fig. 2.

Survival plot for Cox regression. a) CLD increases incident CKD risk; b) CKD increases CLD risk. Adjusted for age, sex, marriage, education, sleep duration, smoking, drinking, hypertension, hyperlipemia, diabetes, heart disease, body mass index, and baseline eGFRcysc.

Subgroup analyses were conducted to evaluate whether the associations between CLD and incident CKD, as well as CKD and incident CLD, were modified by key covariates, including age, sex, diabetes, hypertension, and other baseline characteristics (Fig. 3).

Fig. 3.

Forest plots for subgroup analysis. a) CLD increases incident CKD risk; b) CKD increases CLD risk.

* no. of events / total no. ( %).

** adjusted for age, sex, marriage, education, sleep duration, smoking, drinking, hypertension, hyperlipemia, diabetes, heart disease, body mass index, and baseline eGFR.

For the association between CLD and incident CKD, the increased risk was consistent across all subgroups. Notably, the association was stronger among current drinkers (HR=3.36; 95 % CI: 1.92–5.88) compared to past (HR=2.12; 95 % CI: 0.84–5.39) and never drinker (HR: 1.28; 95 % CI: 0.75–2.16) (Pfor interaction: 0.039). No significant interactions were observed for age, sex, or other group (Pfor interaction > 0.05), indicating that the association between CLD and CKD was robust across these subgroups.

For the association between CKD and incident CLD, the increased risk was also observed across all subgroups. The association was more pronounced in age < 60 years participants (HR=2.18; 95 % CI: 1.63–3.18) compared to those age ≥ 60 (HR=1.45; 95 % CI: 1.00–2.11) (Pfor interaction: 0.003). Like the CLD-CKD analysis, no significant interactions were found for other groups (Pfor interaction > 0.05).

This longitudinal study, based on data from the CHARLS from 2011 to 2020, provides robust evidence of a bidirectional association between CLD and CKD. Participants with baseline CLD had a significantly increased risk of developing CKD, while those with baseline CKD were at a higher risk of developing CLD. These associations remained consistent across sensitivity analyses and subgroup analyses, suggesting that the relationship between CLD and CKD is independent of traditional risk factors such as age, sex, diabetes, and hypertension. The findings highlight the importance of integrated management strategies targeting both liver and kidney health, particularly in high-risk populations.

Our findings align with and extend previous research on the relationship between liver and kidney diseases. Several cross-sectional studies have reported associations between liver dysfunction and reduced kidney function. For example, Mantovani et al. demonstrated that non-alcoholic fatty liver disease was independently associated with an increased risk of CKD in a meta-analysis of observational studies [19]. Similarly, Musso et al. found that NAFLD was linked to a higher prevalence of CKD, even after adjusting for metabolic risk factors [30]. However, most of these studies were cross-sectional, limiting their ability to establish temporal relationships. Our study addresses this gap by using longitudinal data to demonstrate a bidirectional relationship between CLD and CKD.

Conversely, fewer studies have explored the impact of CKD on liver disease. A study by Tonon et al. reported that kidney injury was associated with worse outcomes in patients with cirrhosis, including higher rates of liver-related complications [31]. Our findings build on this evidence by showing that CKD not only exacerbates existing liver disease but also increases the risk of incident CLD in a general population. This bidirectional relationship underscores the interconnected nature of liver and kidney pathophysiology.

The bidirectional association between CLD and CKD may be explained by several shared pathophysiological mechanisms. Chronic inflammation is a key driver of both conditions. In CLD, systemic inflammation resulting from liver injury can lead to endothelial dysfunction, oxidative stress, and activation of the renin-angiotensin-aldosterone system (RAAS), all of which contribute to kidney damage [32]. Similarly, CKD is characterized by chronic low-grade inflammation and uremic toxin accumulation, which can impair liver function and promote fibrosis [33].

Metabolic disorders, such as diabetes, obesity, and hypertension, are common risk factors for both CLD and CKD. Insulin resistance and hyperglycemia can exacerbate liver steatosis and fibrosis while also accelerating kidney damage through glomerular hyperfiltration and vascular injury [5]. Furthermore, RAAS activation, a hallmark of CKD, can worsen liver fibrosis by promoting hepatic stellate cell activation and collagen deposition [34].

Another potential mechanism is gut dysbiosis, which has been implicated in both liver and kidney diseases. Alterations in gut microbiota can lead to increased intestinal permeability, allowing bacterial endotoxins to enter the circulation and trigger systemic inflammation. This “gut-liver-kidney axis” may play a critical role in the progression of both CLD and CKD [34,35].

The strength of this study is that it utilized a large, nationally representative cohort with a long follow-up period, allowing for robust estimation of temporal relationships between CLD and CKD. However, several limitations should be acknowledged. A key limitation is that the definition of chronic liver disease in this study was based on self-reported physician diagnosis from the CHARLS survey and did not distinguish specific etiologies such as NAFLD or MAFLD/MASLD. Therefore, our findings may not fully capture the spectrum or recent nomenclature updates of chronic liver diseases. Second, while eGFRcysc were used to define CKD, the lack of repeated measurements may have led to misclassification of transient kidney dysfunction as CKD. Moreover, lack of information on proteinuria may underestimate the prevalence of CKD. Third, the observational nature of the study precludes causal inference, and unmeasured confounders, such as genetic predisposition or environmental exposures, may have influenced the results. Finally, the findings may not be generalizable to populations outside of China, given the unique demographic and epidemiological characteristics of the CHARLS cohort.

The bidirectional relationship between CLD and CKD has important implications for clinical practice. First, clinicians should be aware of the increased risk of CKD in patients with CLD and vice versa. Routine screening for kidney function in patients with liver disease and liver function in patients with CKD may facilitate early detection and intervention. Second, integrated management strategies targeting shared risk factors, such as diabetes, hypertension, and obesity, are essential to prevent the progression of both conditions. Third, the findings highlight the need for multidisciplinary care involving hepatologists, nephrologists, and primary care providers to optimize outcomes for patients with coexisting liver and kidney diseases. Recent evidence has demonstrated that the coexistence of chronic liver disease and chronic kidney disease further increases the risk of major adverse cardiovascular events (MACE), beyond the risk conferred by each condition alone [36,37]. This synergistic effect is likely driven by the overlapping burden of metabolic dysfunction, systemic inflammation, and vascular injury inherent to both diseases. Thus, clinicians should be aware not only of the reciprocal relationship between CLD and CKD, but also of their combined impact on cardiovascular outcomes.

Future research should aim to address the limitations of this study and further elucidate the mechanisms underlying the bidirectional relationship between CLD and CKD. Longitudinal studies with repeated measurements of liver and kidney function are needed to better characterize the temporal dynamics of disease progression. Additionally, studies exploring the role of gut microbiota, systemic inflammation, and metabolic pathways in the liver-kidney axis could provide valuable insights into potential therapeutic targets. Randomized controlled trials evaluating the efficacy of interventions, such as RAAS inhibitors, anti-inflammatory agents, or lifestyle modifications, in reducing the risk of both CLD and CKD are also warranted. Finally, research in diverse populations is needed to determine the generalizability of these findings and to identify population-specific risk factors and interventions.