Iron and ferritin effects on intensive care unit mortality: A meta-analysis

Abstract

BACKGROUND

The effect of serum iron or ferritin parameters on mortality among critically ill patients is not well characterized.

AIM

To determine the association between serum iron or ferritin parameters and mortality among critically ill patients.

METHODS

Web of Science, Embase, PubMed, and Cochrane Library databases were searched for studies on serum iron or ferritin parameters and mortality among critically ill patients. Two reviewers independently assessed, selected, and abstracted data from studies reporting on serum iron or ferritin parameters and mortality among critically ill patients. Data on serum iron or ferritin levels, mortality, and demographics were extracted.

RESULTS

Nineteen studies comprising 125490 patients were eligible for inclusion. We observed a slight negative effect of serum ferritin on mortality in the United States population [relative risk (RR) 1.002; 95%CI: 1.002-1.004). In patients with sepsis, serum iron had a significant negative effect on mortality (RR = 1.567; 95%CI: 1.208-1.925).

CONCLUSION

This systematic review presents evidence of a negative correlation between serum iron levels and mortality among patients with sepsis. Furthermore, it reveals a minor yet adverse impact of serum ferritin on mortality among the United States population.

Key Words: Iron; Ferritin; Mortality; Critically ill; Meta analysis

Core Tip: This systematic review presents evidence of a negative correlation between serum iron levels and mortality among patients with sepsis. Furthermore, it reveals a minor yet adverse impact of serum ferritin on mortality among the United States population. This guide provides direction for future prognostic assessments in patients with sepsis. Further high-quality cohort studies and experimental studies on molecular mechanisms are needed to confirm our findings.

INTRODUCTION

Iron is an essential nutritional element of the human body with many important biological functions[1]. With the varying cellular environment, the interconversion between the two oxidized states of iron (Fe²⁺ and Fe³⁺) keeps iron relevant to a variety of important biochemical reactions, but also with potential hazard risks[2]. Under oxygenic conditions, the Fenton/Haber-Weiss reaction of free iron catalyzes the production of harmful atomic groups, and this mechanism has been hypothesized to underlie iron toxicity in some pathological states[3].

With the progress of clinical and experimental studies, iron metabolism has been shown to play a crucial role not only in metabolic diseases, but also in monitoring disease prognosis using serum iron indicators[4-6]. Changes in iron metabolism indicators occur in critically ill patients, and some of these serum iron metabolism indicators have been reported to be used to predict prognosis in these patients. In a study on multiple serum iron metabolism indicators among 51 critically ill patients after surgery, elevated serum ferritin was associated with poor clinical outcomes[7]. Another study on a prospective cohort of 121 patients under critical care in an integrated intensive care unit (ICU) revealed a significant association between plasma iron levels and increased risk of inpatient 30-d mortality[8]. According to another prospective study, serum iron metabolism indicators are associated with prognosis of ICU patients. In-survival analysis revealed that lower serum iron and iron utilization levels could increase short-term and long-term survival outcomes. Furthermore, after multivariate analysis adjusting for other indicators, iron was found to be an independent predictor of short-term and long-term survival outcomes of mortality among ICU patients[9]. Currently, no conclusions have been established as to whether serum iron metabolism indicators can accurately predict the prognosis of critically ill patients. The relevant underlying mechanisms are still being explored.

Therefore, a better understanding of the importance of iron metabolism among ICU patients is urgently needed. In consequence, we analyzed the effect of iron parameters on mortality among critically ill patients admitted to ICU.

MATERIALS AND METHODS

Search strategy

This meta-analysis study was conducted by following the Meta-analysis of Observational Studies in Epidemiology guidelines[10]. A literature search of relevant published studies that analyzed the association between iron parameters and mortality outcomes among critically ill patients was conducted on October 1, 2021. Using literature searches on PubMed, Embase, Web of Science, and Cochrane Library databases, we identified articles using the following terms: Ferritin, iron, anemia, critically ill, sepsis, sepsis shock, and mortality. Only studies published in English were identified for the analysis in this study. By consulting the reference lists in these research articles, we identified additional studies relevant to the subject.

Study selection criteria

Studies were included, if: (1) They were cohort or case-control studies; (2) the study evaluated the association between iron parameters and mortality among critically ill patients admitted to ICUs; (3) they provided sufficient data on odds ratios (OR); and (4) the Newcastle-Ottawa scale (NOS) score was ≥ 6. All studies containing overlapping data were excluded.

The data extracted included the first author’s name, year of publication, study population and country, period, study type, age, ORs with 95%CI, ICU type, patient types, and NOS scores (Table 1). For publications that did not report the association between iron or ferritin and mortality among critically ill patients, we calculated the ORs, if data on these variables were available.

Table 1 Characteristics of the studies included in the meta-analysis.

Ref.	Population and country	Period	Study type	Age	Adjustment OR (95%CI)	ICU type	Patient types	NOS
Rimmelé et al[13]	1260, Italy	From February 22 to May 31, 2020	Cohort	Median age of 63 yr	Ferritin: 1.000 (1.000–1.000)	ICUs	COVID-19	7
Deng et al[14]	100, China	From January 30 to March 30 2020	Cohort	Median age of 65 yr	Ferritin: 32.63 (8.30–128.32)	ICUs	COVID-19	7
Lachmann et al[17]	116310, Germany	From January 2006 to August 2018	Cohort	18 years old or older	Ferritin: 1.518 (1.384–1.665)	Adult ICU	Critically ill patients	8
Tonial et al[15]	1407, Brazil	From July 2013 to January 2017	Cohort	6 months to 18 yr old	Ferritin: 5.17 (2.619-10.205)	Pediatric ICU	Sepsis	8
Brandtner et al[18]	61, Austria	From February 2018 to December 2019	Cohort	Mean age of 63 yr	Ferritin: 6 (1.1-39)	Adult ICU	Sepsis	7
					Iron: 4.6 (1.5-17)
Shu et al[16]	483, United States	From 2001 to 2012	Cohort	Aged 16 yr or above	Ferritin (90 D): 1.001 (1.000-1.001)	Adult ICU	AKI	6
					Iron (90 D): 1.741 (1.285-2.358)
Jiang et al[20]	258, China	From May 1 2016 to November 30 2017	Cohort	18 yr old or older	Ferritin: 1.010 (1.000–1.021)	Adult ICU	Sepsis	6
Xia et al[19]	250, China	From June 2015 to May 2017	Cohort	18 yr old or older	Iron: 0.696 (0.268-0.918)	Adult ICU	Critically ill patients	6
Lan et al[23]	1891, United States	From 2001 to 2012	Cohort	Aged 16 yr or above	Ferritin: 1.000 (0.999–1.000)	Adult ICU	Sepsis	6
					Iron: 1.506 (1.190–1.908)
Xie et al[21]	103, China	From August 2016 to January 2017	Cohort	18 yr old or older	Ferritin (long-term): 1.002 (1.000–1.003)	NCU	Neuro-critically patients	6
Lasocki et al[22]	2087, France	From August 2011 to June 2013	Cohort	Unlimited	Ferritin: 1.02 (0.51-2.06)	ICUs	Iron deficiency	7
Ghosh et al[24]	42, South Asia	From May 2016 to March 2017	Cohort	6 months to 12 yr old	Ferritin: 1.94 (0.94–4.02)	Pediatric ICU	Septic shock	6
Tacke et al[9]	312, Germany	Not provided	Cohort	Median age of 64 yr	Ferritin (long-term): 1.000 (1.000–1.000)	Adult ICU	Critically ill patients	7
					Iron (long-term): 1.040 (1.011–1.071)
Unal et al[26]	111, Turkey	Between May 2012 and January 2013	Cohort	Mean age of 73.79	Ferritin: 1.002 (1.000-1.004)	Adult ICU	Critically ill patients	6
Carcillo et al[29]	100, Pittsburgh	Not provided	Cohort	Age greater than or equal to 44 wk gestation and less than 18 yr	Ferritin: 1.001 (1.000–1.001)	Pediatric ICU	Sepsis	7
Uscinska et al[25]	392, Poland	Between 2008 and 2011	Cohort	Mean age of 70	Iron: 0.85 (0.78–0.94)	ICUs	Anemia	7
Leaf et al[8]	121, United States	Between 2008 and 2012	Cohort	Age >18 yr	Iron (long-term): 1.38 (0.91-2.10)	ICUs	AKI	6
Bennett et al[27]	171, United States	Between September 2, 2003, and August 16, 2007	Cohort	Age < 18 yr	Ferritin: 4.32 (2.21-8.47)	Pediatric ICU	Critically ill patients	6
Garcia et al[28]	31, Brazil	From January 2004 to September 2005	Cohort	Aged 1 month–16 yr	Ferritin: 3.2 (1.3-7.9)	Pediatric ICU	Sepsis	6

OR: Odds ratio; NOS: Newcastle-Ottawa scale; ICU: Intensive care unit; NCU: Neuro critical care unit; COVID-19: Coronavirus disease 2019; AKI: Acute kidney injury.

Statistical analysis

The strength of the association between iron or ferritin and mortality among critically ill patients was reported using ORs and 95%CIs. When adjusted and crude ORs were provided, the most adjusted ORs were extracted. We used the heterogeneity value (I²) test and Q-statistic to detect any possible heterogeneity, as a quantitative measure of inconsistency among studies[11]. Meta-regression analyses were used to investigate the sources of heterogeneity. Pooled ORs and 95%CIs were calculated using a random-effects model[12].

All statistical analyses in the meta-analysis were performed using Statistical software for data science version 13.0. All reported P values were from two-sided statistical tests. Statistical significance was set at P ≤ 0.05. Egger’s and Begg’s regression models were used to evaluate potential publication bias[11].

RESULTS

The process for selecting the studies is outlined in Figure 1. After the literature search, 267 potentially relevant records were reviewed. Out of this number, 19 studies, including 125490 patients, were included in the meta-analysis (Table 1). Subsequently, 247 studies were excluded, because they were either duplicated reports or were of relatively low quality. All 19 selected studies were cohort studies[8,9,13-29].

Figure 1 Search strategy and selection of studies for inclusion in the meta-analysis.

Six studies were conducted in Europe; five studies were conducted in America; and five studies were conducted in other regions (Asia, Australia). Four studies presented data on mortality and iron and ferritin separately. The NOS scores of eight, six, and two studies were 6, 7, and 8, respectively (Table 2). The data on mortality and serum iron and ferritin were extracted separately. Most of the included studies had contrasting findings. Six studies reported that high ferritin levels were associated with high mortality among critically ill patients, whereas the other 10 studies reported no association. Similarly, the studies on serum iron had contrasting findings. Four studies reported that high serum iron levels were associated with high mortality among critically ill patients, whereas the other three studies reported no association or corresponding decrease in mortality among critically ill patients. The analysis of the studies yielded a combined risk estimate, relative risk (RR) of 1.00 (95%CI: 1.00–1.00; P = 0.001) with I² of 84.4% for ferritin (Figure 2A), and a combined risk estimate, RR of 1.02 (95%CI: 0.99–1.05; P < 0.001) with I² of 85.3% for iron (Figure 2B). We also assessed the stability and explored the sources of heterogeneity of the results using sensitivity analysis for serum ferritin and iron (Figure 3). After a meta-regulation test, geographical area was associated with 50.51% heterogeneity reduction across the studies for serum ferritin and 41.53% heterogeneity reduction across the studies for serum iron (Figure 4).

Figure 2 Meta-analysis between ferritin and mortality in critically ill patients. A: Ferritin; B: Iron.

Figure 3 Sensitivity analysis of all included studies. A: Ferritin; B: Iron.

Figure 4 Meta-regulation of ferritin and mortality in critically ill patients. Geographical area, 1: Europe, 2: Others, 3: America. A: Ferritin; B: Iron.

Table 2 Stratified analysis of iron parameters and mortality in critically ill patients according to study characteristics.

Group	Number of the study	RR (95%CI)	P value	I2 (%)
Ferritin
Geographic region
Europe	6	1.001 (1.001-1.002)	< 0.01	94.3
America	5	1.002 (1.002-1.004)	0.293	19.1
Others	5	1.001 (1-1.001)	0.001	78.2
NOS
6	8	1.001 (1.001-1.001)	0.001	70.5
7	6	1.001 (1.001-1.001)	0.72	0
8	2	1.523 (1.383-1.663)	0.059	71.9
Patient types
Sepsis	7	1.001 (1.001-1.001)	0.004	68.2
Other critically ill patients	9	1.001 (1.001-1.002)	< 0.01	89.9
ICU types
Adult ICU	8	1.001 (1-1.001)	0	91.3
Pediatric ICU	5	1.001 (1.001-1.002)	0.017	66.9
General ICU	3	1.025 (0.25-1.8)	0.302	6.2
Iron
Geographic region
Europe	2	1.017 (0.988-1.045)	< 0.01	94.7
America	3	0.858 (0.573-1.143)	0.09	58.4
Others	2	1.616 (1.318-1.914)	0.583	0
NOS
6	4	1.217 (1.011-1.424)	0.001	82.7
7	3	1.017 (0.989-1.045)	< 0.01	89.9
Patient types
Sepsis	2	1.567 (1.208-1.925)	0.442	0
Other critically ill patients	5	1.017 (0.989-1.045)	< 0.01	87.2
ICU types
Adult ICU	5	1.043 (1.013-1.073)	0.001	79.7
General ICU	2	0.859 (0.780-0.939)	0.084	66.6

RR: Relative risk; NOS: Newcastle-Ottawa scale; ICU: Intensive care unit; I²: I-squared statistic.

Due to the differences in geographic area (Europe, America, and others), NOS (6, 7, or 8), patient category (sepsis, other critically ill patients), and ICU type [adult ICU, pediatric ICU (PICU), or general ICU] among the studies, we further conducted subgroup analyses to determine the effect of these factors in our analyses (Table 2). We found a significant negative effect of serum ferritin on mortality in the United States population (RR = 1.002; 95%CI: 1.002-1.004) and in the general ICU (RR = 1.025; 95%CI: 0.25-1.8). We obtained a statistically negative effect on patients with sepsis (RR = 1.567; 95%CI: 1.208-1.925) for serum iron.

Based on Egger’s and Begg’s regression models, evidence of publication bias (Figures 5 and 6) were noted for iron or ferritin, in relation to mortality. Egger’s funnel plot and Begg’s linear regression test revealed P values < 0.05.

Figure 5 Begg’s funnel plot assessing publication bias among the selected studies. A: Ferritin; B: Iron.

Figure 6 Egger’s funnel plot assessing publication bias among the selected studies. A: Ferritin; B: Iron.

DISCUSSION

To the best of our knowledge, this study is the first systematic review examining the role of iron parameters on mortality in critically ill patients. High iron parameters had no association with mortality among critically ill patients. Due to the heterogeneity of the meta-analysis, we performed subgroup analysis, sensitivity analysis, and regression analysis to explore the source of heterogeneity. Subgroup analyses were used to determine the effect of geographic area (Europe, America, or others), NOS score (6, 7, or 8), patient types (sepsis, other critically ill patients), and ICU types (Adult ICU, PICU, or general ICU) in our analyses (Table 2). For iron, sepsis had a statistically negative effect (RR = 1.567; 95%CI: 1.208-1.925). Sensitivity analysis revealed stable conclusions and no apparent source of heterogeneity for serum ferritin and iron (Figure 3). From the meta-regulation test, geographical area was associated with 50.51% heterogeneity reduction across studies on serum ferritin and 41.53% heterogeneity reduction across studies on serum iron (Figure 4). Due to publication bias, further studies of higher quality are needed to confirm our findings.

The meta-analysis concluded that serum ferritin had no association with mortality among critically ill patients. However, since few studies with high level of heterogeneity were included, exact conclusions could not be made. Ferritin is an indicator of iron level and storage in the body. However, it is also an acute phase reactive protein. A significantly increased serum ferritin level in stress conditions, such as inflammation, does not accurately reflect the body's iron storage. Pro-inflammatory cytokines can induce increased ferritin transcription and translation, resulting in elevated serum ferritin level. Inflammation is a common pathological state in critically ill patients that may likely result in multi-organ dysfunction, depending on disease severity[30,31]. According to studies on iron parameter levels in critically ill patients, serum ferritin levels in critically ill patients are significantly higher during sepsis than during the recovery stage. This implies that serum ferritin level can reflect the state and severity of infection[7]. Additionally, inflammatory stimuli can upregulate the expression of ferromodulin and modulate iron metabolism by inhibiting intestinal iron absorption and limiting iron release in the mesh endothelial system, resulting in a relative iron deficiency state in the circulatory system[32-34]. Currently, the nature of association between ferritin level and mortality among critically ill patients is yet to be elucidated. To remedy the situation, separate clinical studies on different diseases with varying severities in different populations should be conducted.

The meta-analysis concluded that no association existed between iron level and mortality in critically ill patients, although subgroup analysis revealed that in patients with sepsis, iron is a risk factor. Under normal physiological conditions, iron metabolism is in a balanced state, which does not only maintain the important biological function of iron but also prevents excessive iron accumulation that could result in oxidative stress due to injury to the body. Iron homeostasis ensures that changes in iron metabolism present in ICU patients are essentially a defense mechanism. From initial studies, the trend of serum iron metabolism indicators in critically ill patients was similar to that in chronic inflammation, which is a common pathological state in critically ill patients. Infectious inflammation is the most common pathological state[35,36]. One feasible mechanism underlying the high iron level is the state of histiocyte breakdown in ICU patients. The disruption of erythrocytes and other tissues gradually decreases iron consumption and increases iron release. As a result, repeated blood transfusions further increases iron level in circulation after erythrocyte lysis[37-39]. The increase in iron-associated mortality may be due to excess iron, which directly promotes increased infection in patients with sepsis. Furthermore, iron catalyzes the chemical production of reactive oxygen species, such as hydroxyl anions and superoxide, thereby accelerating the development of multi-organ dysfunction[40]. There is a lack of definitive evidence on the effect of iron on survival and mortality in critically ill patients, except those with sepsis.

This study had few limitations. First, only studies published in English journals were included. However, a significant portion of the literature were studies conducted in Asia, where the official language is not English. Second, predicting the effect of misclassification of cohort studies on the results was challenging. Third, systematic confounding or the risk of bias in observation studies cannot be ruled out easily. Fourth, due to heterogeneity across the studies, regression analysis was used to explain the source of heterogeneity, which may be due to differences in study geographical areas. In this study, analyses of the association between concentration of ferritin and duration of ferritin abnormality and mortality among critically ill patients were not included. The original studies lacked data on these parameters for this analysis.

CONCLUSION

In summary, this systematic review presents evidence of a negative correlation between serum iron levels and mortality among patients with sepsis. Additionally, it identifies a minor but adverse effect of serum ferritin on mortality within the United States population. Further high-quality cohort studies and experimental studies on molecular mechanisms are needed to confirm our findings.