Chronic kidney disease (CKD) is characterized by a progressive loss of renal structure and function, which may eventually lead to irreversible renal impairment. The global prevalence of adult CKD ranges from approximately 15% to 20%, with the prevalence of adult CKD in mainland China estimated at approximately 8.2% (1–3). Patients with end-stage renal disease (ESRD) must rely on renal replacement therapy (RRT), such as dialysis or renal transplantation, in order to continue to live. It is possible that those affected may experience a greater economic burden and a lower quality of life (4, 5). Therefore, the development of efficacious interventions to impede the progression of CKD has become a current research priority in the field of nephrology.

Anemia is a prevalent complication in patients with CKD, with a twofold higher prevalence than in the general population (6). Impaired renal function results in a microinflammatory state, decreased oxidative capacity, and abnormal endocrine function. Together, these factors lead to decreased erythropoietin (EPO) production and abnormal absorption and transport of hematopoietic raw materials such as iron. Research indicates that anemia has a detrimental impact on the quality of life of patients with CKD. On the one hand, anemia is associated with adverse symptoms such as fatigue and depression. On the other hand, anemia increases the risk of entering ESRD and developing all-cause mortality (6–9). The prompt intervention and management of anemia is of significant importance in the slowing down and prevention of the progression of CKD.

Hemoglobin (Hb) and hematocrit (HCT) are commonly employed in the assessment and management of anemia, as they provide a direct reflection of the status and function of red blood cells. Hb is responsible for transporting oxygen and is directly related to the body’s oxygen supply, which is a key substance for maintaining normal physiological functions of the body. HCT is the proportion of red blood cells and is crucial for assessing the status of red blood cell production and destruction. Therefore, Hb and HCT play an important role in the diagnosis, treatment, and management of renal anemia. Hb and HCT can be effectively improved by conventional pharmacological means of treatment, including EPO, iron, and folic acid, as well as the latest drugs such as roxadustat (10, 11).

However, the optimal range for regulating levels of Hb and HCT remains uncertain, with conflicting recommendations present across guidelines. For instance, the 2012 the Kidney Disease: Improving Global Outcomes (KDIGO) guideline recommended that the Hb of patients with CKD should be controlled at 110–120 grams per liter (g/L), corresponding to an HCT of 33%–36%. In contrast, the 2017 UK NICE guideline recommended controlling the level of Hb of adults at 100–120 g/L. Researchers in mainland China proposed that the optimal therapeutic target for Hb in patients with CKD should be within the range of 115–130 g/L. It was further recommended that Hb levels should be individually adjusted within the range of 110–120 g/L in accordance with the patient’s chronological age, the mode and duration of dialysis, the duration of the erythropoiesis-stimulating agent (ESA) treatment, and the presence of other concomitant diseases (12–14). Furthermore, there are only a few guidelines that suggest a growth rate in Hb during anemia treatment. For instance, the Chinese expert consensus on renal anemia states that the initial goal of ESA treatment is to increase Hb by 10–20 g/L per month, and to avoid an increase in Hb of more than 20 g/L within 1 month (14). The controversy surrounding the Hb targets makes it challenging for physicians to identify a clear standard in the actual treatment process, which in turn increases the complexity and uncertainty of anemia management.

The optimal range for level of Hb is a matter of contention for a number of reasons. Firstly, the majority of studies only observe the relationship between baseline Hb level and disease prognosis, whereas in reality, the Hb concentration undergoes changes with the CKD progression. Focusing solely on a specific point in time is not sufficiently objective to analyze the relationship between anemia and disease prognosis (8, 9, 15). Furthermore, most studies on the prognosis of patients with CKD with anemia have employed stratified comparative analyses, which have yielded disparate conclusions, leading to heterogeneity in Hb control criteria (16, 17). Thirdly, a number of pertinent studies were conducted among Western populations, with fewer studies conducted in Asian populations (16, 17).

In light of the controversies previously outlined, our study investigates the relationship between baseline level and longitudinal change of Hb and HCT and CKD prognosis in Chinese patients with CKD stages 3–4 in order to elucidate the intrinsic link between Hb and HCT and CKD short- and long-term prognosis, as well as to determine the optimal range of Hb and HCT control.

We retrospectively collected demographic and clinical information over a 6-year follow-up period to investigate the association between Hb and HCT and the prognosis of patients with CKD stages 3–4 in Lingnan, China. The final cohort comprised 730 patients with CKD stages 3–4 in accordance with the inclusion and exclusion criteria. The baseline Hb level that met the recommended range of the Chinese renal anemia guideline was approximately 60%–63%, which was comparable to the Hb attainment rate of Japanese patients with CKD stages 3–4 and slightly higher than that of patients with CKD stages 3–4 in the United States (20, 21). Moreover, the proportion of individuals in the substandard group exhibited a decline from 40.0% to 14.8% over the course of the follow-up period, which may be related to implementation of effective anemia management strategies and a higher incidence of the composite outcome among the substandard group.

We constructed a time-dependent Cox model with baseline and first-year mean values of Hb and HCT as independent variables, and a GBTM model with longitudinal measurements of Hb and HCT, respectively, to explore the association of Hb and HCT with the short- and long-term risk of composite outcomes of patients with CKD stages 3–4. The results of the time-dependent Cox model showed that after adjusted for covariates, the mean_Hb [HR (95% CI): 0.851 (0.786, 0.921) g/L, p=0.000] and the mean_HCT [HR (95% CI): 0. 578 (0.441, 0.758) %, p=0.000] were the independent protective factors and baseline_Hb and baseline_HCT were not independent factors associated with the composite outcome. It reveals a closed relationship between mean_Hb and mean_HCT and CKD prognosis, and on the other hand, it suggests that levels of Hb and HCT may vary with the progression of the disease and the different treatments employed in patients with CKD stages 3–4. While baseline values remain clinically important for initial assessment and treatment initiation, our findings suggest that longitudinal monitoring of Hb and HCT levels provides superior prognostic information compared to single-point measurements. The findings indicate that patients with CKD require effective management and maintenance of their levels of Hb and HCT over time to improve their prognosis.

To gain further insight into the management of long-term anemia in patients with CKD, we constructed a GBTM using longitudinal data. This approach enabled us to divide the persistence values of Hb and HCT in these patients into four distinct groups based on their trajectory. We then compared the prognosis of these groups. Taking group 1 (“lower and decreasing”; Hb<100 g/L, HCT approximately 30%) as reference, group 2 (“lower and growing slightly”; Hb 110–120 g/L, HCT approximately 35%), group 3 (“higher and growing slightly”; Hb 125–135 g/L, HCT approximately 40%), and group 4 (“higher and growing steadily”; Hb 145–160 g/L, HCT approximately 45%) were independent protective factors for the occurrence of the composite outcome in patients with CKD stages 3–4 (p=0. 000). Group 1 (“lower and decreasing”; Hb<100 g/L, HCT approximately 30%) had more fluctuations than groups 2 to 4, which were relatively stable. As Hb and HCT moved from a lower to a higher level, the HRs decreased gradually, indicating a gradual decreased risk of composite outcomes in patients with CKD stages 3–4 (p for trend<0.000). We further compared baseline characteristics between groups 4 and 1. Compared with group 1, group 4 exhibited fewer comorbidities, a lower proportion of certain medications, and a more favorable physiologic and metabolic profile ( Supplementary Data Sheet 4 ). These findings suggest that younger patients with fewer comorbidities may better tolerate higher Hb levels, and that preserved kidney function and better nutritional status may attenuate the risks associated with higher Hb.

The results of both the time-dependent Cox model and the GBTM model reveals that higher levels of Hb and HCT had a lower risk of short- and long-term disease progression in CKD. This finding is broadly consistent with previous studies in Asian populations (15). Previous studies have postulated that the observed improvement in hematopoiesis may be attributed to a reduction in metabolic pressure in the proximal tubules or adjacent interstitium of the kidney during the treatment process. This reduction in pressure is hypothesized to enhance the production of red blood cells, as evidenced by elevated levels of Hb and HCT. Therefore, higher levels of Hb and HCT can be considered a surrogate indicator of improved renal function (22). Subgroup analyses showed an interaction effect of mean_Hb with the subgroup of sex (p for interaction=0.034) and an interaction effect of group 2 of Hb trajectory with the subgroup of CKD stage (p for interaction=0.015). It suggests that the effect of Hb and HCT on the composite outcome may vary by sex and CKD stages, especially those with Hb levels ranging from 110 to 120 g/L.

We collected clinical data of patients with non-dialysis CKD based on a retrospective cohort study, attributing to a limited accuracy of determinations of cardiovascular outcomes. In view of this, we did not further investigate the association between Hb and HCT and cardiovascular outcomes in patients with CKD. However, it is worth noting that previous studies have shown that too high a level of Hb may be associated with an increased risk of cardiovascular events. For example, The Correction of Hemoglobin and Outcomes in Renal Insufficiency (CHOIR) trial included 1,432 patients with CKD stages 3–4 with anemia (defined as Hb <110 g/L) and randomly arranged them to groups with a higher or lower treatment target (135 g/L vs. 113 g/L) with a study duration of 3 years. It showed that the group with a higher treatment target had a higher HR (95% CI) of 1.34 (1.03,1.74) (p=0.03) after the occurrence of the first cardiovascular events compared to the group with a lower treatment target and no significant gain in quality of life (16). Some researchers had doubts about the conclusion because the group with the higher treatment target had a higher proportion of history of hypertension and history of undergoing coronary artery bypass grafting and a greater severity of congestive heart failure than the group with the lower treatment target at baseline (23). After adjusting for heart failure scores at baseline, the correlation between the two groups and the primary composite outcome was not significantly different [HR (95% CI) 1.24 (0.95, 1.62), p=0.11] (24). The Cardiovascular Risk Reduction by Early Anemia Treatment with Epoetin beta (CREATE) study obtained an opposite conclusion. It included 603 CKD stages 3b–4 patients with mild to moderate anemia (defined as Hb between 110 and 125 g/L) and also randomized them into groups with a higher or a lower treatment target (130–150 g/L vs. 105–115 g/L) with a study duration of 3 years. It showed that the group with a higher treatment target had a higher HR (95% CI) of 0. 78 (0.53, 1.14) (p=0.20) after the occurrence of the first cardiovascular events and had a higher quality of life than the group with the lower treatment target (17). The Kidney Disease Outcomes Quality Initiative (KDOQI) work group pooled the results of six related studies, and it showed that groups with higher Hb targets yielded a relative risk (RR) (95% CI) of 1.24 (1.02, 1.51) on adverse cardiovascular events in patients with non-dialysis CKD. However, the CHOIR trial and the CREATE study contributed 94% of the weight in this meta-analysis (25).

In light of the previous studies and the findings of our study, it is evident that there is a discernible advantage to maintaining a higher level of Hb and HCT over time in patients with CKD stages 3–4 to thwart the progress of CKD [including all-cause mortality, ESRD (RRT and eGFR <15 mL/min/1. 73m²), doubling of SCr from baseline, and ≥50% reduction in eGFR from baseline]. It should be noted that this benefit is contingent upon maintaining Hb and HCT at a relatively stable level over time, rather than merely reaching a specific point in time. In our study, we also did not find that having excessive levels of Hb and HCT were detrimental to the composite outcome. We think that it is important to note that the level of Hb and cardiovascular events remains controversial. Our retrospective study had limitations in exploring this issue. From a more cautious perspective, we recommend that Hb levels be maintained within the range of 110 to 130 g/L in patients with non-dialysis CKD.

First, although our study collected as much information as possible and used Cox regression and GBTM methods to adjust for confounding factors, it is important to note that the retrospective study design itself still has limitations in terms of study design and data collection. There are still potential confounding factors that cannot be fully adjusted. It should be noted that, because this is a retrospective study and we lacked data on cumulative ESA or iron doses and dosing frequency, residual confounding related to treatment intensity and medication response may persist. Additionally, as in many retrospective studies, selection bias is possible—particularly if less healthy patients are more likely to be lost to follow-up. Consistent with this concern, our baseline comparisons indicated potential selection bias: the included cohort was younger; had lower eGFR, Hb, and HCT; had higher urea and slightly higher ALT; had more frequent use of sodium bicarbonate and potassium-lowering agents but less calcium supplementation; and had a longer follow-up with fewer composite outcomes than the excluded cohort ( Supplementary Data Sheet 1 ). Second, Hb and HCT are subject to short−term biological and pre−analytical variability (e.g., diurnal rhythms, hydration, and volume status), which may introduce minor bias. To mitigate this, baseline laboratories were obtained primarily under standardized procedures at GPHCM, and we adjusted for covariates related to volume and nutrition (e.g., eGFR, serum albumin, use of diuretics and ACEI/ARB, and presence of hypertension). Residual short-term variability may remain, but these steps reduce its impact. In addition, despite aggregating measurements into 3-month windows for GBTM, some trajectory misclassification related to irregular measurement timing may persist. Third, although the association between Hb level and risk of cardiovascular events remains unclear, there is a theoretical possibility of competing risks of cardiovascular outcomes in our study, and analysis of this potential risk was not possible given the limitations of the data. Fourth, the size of our study sample is relatively limited, and further increases in sample size, improvements in study methodology, or additional investigations of associated factors are required to enhance our understanding of the pathogenesis of the disease and predict future outcomes for patients. Fifth, it should be noted that our study only includes data from southern China, and its generalizability needs to be considered.