Beyond anemia: a comprehensive analysis of iron deficiency symptoms in women and their correlation with biomarkers

Abstract

Background

Iron deficiency, the most widespread micronutrient deficiency globally, is especially common among women and seriously impacts their health considerably. The symptom spectrum in non-anemic iron deficiency (NAID) is less well described than in anemic forms. The aim of this study was to demonstrate the diversity and frequency of symptoms in NAID and compare with anemic cases and explore the association of symptoms with biomarkers.

Methods

This retrospective and cross-sectional study was conducted on female patients at a university hospital. Patients admitted to the iron deficiency outpatient clinic between March 2021 and August 2022 were analyzed. Patients aged 18 years and older with iron deficiency (ferritin < 30 mcg/L) were included. Iron parameters, hemoglobin levels, and symptoms were included into the dataset, and these independent variables were compared between patient groups categorized as NAID and iron deficiency anemia (IDA).

Results

The study sample consisted of 239 patients. The frequencies of IDA and NAID were 70.3% (n = 168) and 29.7% (n = 71), respectively. Forty-one distinct symptoms or symptom groups were documented, ranging from neuropsychiatric to gastrointestinal manifestations. The mean number of symptoms per person was 16.5 (7.03). The top ten most common symptoms were weakness (87.03%), fatigue (82.43%), easy fatigability (78.66%), amnesia (72.38%), feeling cold (71.55%), alopecia (70.29%), cold intolerance (69.04%), sleep problems (66.53%), nervosity (63.18%), and cold foot (59.83%). The IDA group had more diverse symptoms than the NAID group (P <.05). Most of the forty-one symptoms or symptom groups were similar in frequency in the IDA and NAID groups. Only palpitations, cold intolerance, easy fatigability, and unclassifiable group differed between the groups (P <.05). The highest correlation between iron parameters and symptom diversity was transferrin saturation (TSAT) (r=-.229, P <.001). All but three of the positive symptoms showed TSAT levels below 8%.

Conclusions

This study confirms that anemia is not a prerequisite for the majority of iron deficiency symptoms, which are common in non-anemic women. TSAT demonstrates a robust correlation, underscoring its potential as an early diagnostic biomarker. Clinicians must acknowledge various non-specific symptoms irrespective of anemia. Additional research is required to corroborate findings and investigate symptom mechanisms and treatments.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12905-025-03906-w.

Graphical Abstract

Supplementary Information

The online version contains supplementary material available at 10.1186/s12905-025-03906-w.

Introduction

Iron deficiency is the most common micronutrient deficiency in the world, particularly in women, and is a very common disorder affecting one-third of the world’s population [1]. Another indicator that this disorder is also very common is that iron deficiency ranks ninth in the global burden of disease [2].

The most prominent hematologic consequence of deepening and/or prolonging iron deficiency is iron deficiency anemia (IDA). IDA is an important disease that has been extensively studied, has been shown to impair the function of many organs and has a significant negative impact on the course of other comorbidities.

However, non-anemic or isolated iron deficiency is also an important disorder in its own right. Compared with normal ferritin, non-anemic iron deficiency (NAID) is associated with increased mortality and should therefore be taken at least as seriously as IDA [3].

It has been reported that the economic loss caused by iron deficiency through productivity loss, whether NAID or IDA, is equivalent to 0.57% of the annual gross domestic product [4]. Symptoms that cause a loss of productivity -and a wide range of symptoms associated with iron deficiency- include fatigue, poor sleep, back pain, headaches, insomnia, and anxiety [5].

The low awareness of iron deficiency among clinicians and especially the fact that the adverse outcomes of NAID have not yet been investigated as much as IDA may cause the presence of NAID to be overlooked in patients presenting with nonspecific symptoms [6]. Indeed, certain symptoms have significant diagnostic sensitivity for the diagnosis of iron deficiency [7].

The main hypothesis of this study is to demonstrate that the diversity and prevalence of symptoms in non-anemic iron deficiency are at least as wide as those observed in anemic individuals. Because knowing the range of symptoms of iron deficiency (ID), which is very common, has a significant disease burden and causes serious productivity loss, and determining the relevant symptom rates according to IDA and NAID will make a significant contribution to the awareness of clinicians on this issue. In particular, demonstrating how widespread symptoms are, even in non-anemic iron deficiency, will increase the precision of disease management. On the other hand, examining which basic biomarkers are more strongly associated with symptoms will also contribute to awareness in this direction.

Methods

Study design and patient selection

The population of this cross-sectional and retrospective study consisted of female patients who were admitted to the Ordu University Training and Research Hospital Iron Deficiency Outpatient Clinic between March 2021 and August 2022 and whose iron deficiency status persisted. The main inclusion criteria were female sex, patient age ≥ 18 years and ferritin level below 30 mcg/L. The main exclusion criteria were diabetes mellitus, chronic inflammatory diseases, neuromuscular and neuropsychiatric diseases, cardiopulmonary diseases and other disorders that cause significant chronic symptoms. However, those with widespread data loss were also excluded.

Data type and documenting symptoms

The patients’ age (years), transferrin saturation (%), ferritin (mcg/L), and hemoglobin (g/dL) levels were documented as continuous variables from digital records.

All patients attending the iron deficiency outpatient clinic of our hospital are routinely asked about the presence of nearly forty symptoms. These symptoms are noted as either present or absent. In addition, if the patient had symptoms beyond those typically inquired about at this clinic, those symptoms were also incorporated into the dataset. Consequently, symptom-related data were extracted from the patient’s medical records for the study dataset and documented as categorical data in the form of “present” or “absent.” Moreover, symptoms that exhibit similarities, indicate a specific symptom cluster, or are inadequately characterized by the patient are categorized under a single designation/variable.

The dataset was categorized into two groups according to the existence of anemia: “iron deficiency anemia” (hemoglobin < 12 g/dL) and “non-anemic iron deficiency” (hemoglobin ≥ 12 g/dL) [8].

Statistical analysis

The data were gathered using Microsoft Excel^®, a dataset was generated, and requisite arrangements were implemented. Subsequent to this arrangement, the data were transferred to Jamovi (v2.5.3) and R (v4.4.0) for statistical analysis [9, 10]. The graphs were produced utilizing R (v4.4.0) and Canva (v2) software [10, 11].

Descriptive statistics were first calculated for continuous variables, including age, hemoglobin, iron parameters, and symptoms across the entire sample, presenting mean and standard deviation for normally distributed data, and median and interquartile range (IQR) for non-normally distributed data. Frequency and percentage distributions were provided for categorical variables. Subsequently, these variables were examined within two separate groups (NAID and IDA). For intergroup comparisons of continuous variables, the independent samples t-test or the Mann-Whitney U test was utilized, contingent upon the data distribution. The chi-square test was employed to compare categorical variables. Spearman correlation analysis was performed to evaluate the relationships between variables. A P-value under 0.05 was considered statistically significant.

Results

Descriptive statistics in the entire sample

The number of patients who met the study criteria was 239. The median values of the baseline parameters in the whole sample were as follows: age 39 years (20–66), hemoglobin (HGB) 11.1 g/dL (5.2–14.8), transferrin saturation (TSAT) 7.55% (1.91–55.1) and ferritin (FERR) 7.28 mcg/L (0.5–28.9).

In the records of symptoms, 43 (forty-three) different symptoms were identified. However, the variety of symptoms related to sleep (sleeping less, sleeping more, etc.) were categorized into a single class, and neuropathic complaints (burning, numbness) were categorized into a single class. Unclassifiable complaints were noted as “others”. In the final version, the total number (variety) of symptoms was 41 (forty-one).

The mean number of symptoms per patient was 16.5 (7.03). The symptom types and rates for the whole sample are shown in Fig. 1, with a circular plot drawn with the ggplot2 package of the R program.

Fig. 1.

Circular plot of symptoms and their frequencies in the whole sample. The frequency of each symptom is represented by the length of the bar, with the numerical frequency shown inside the bars

The top ten symptoms and their frequencies are as follows: weakness, 87.03%; fatigue, 82.43%; easy fatigability, 78.66%; amnesia, 72.38%, feeling cold, 71.55%; alopecia, 70.29%; cold intolerance, 69.04%; sleep problems, 66.53%; nervosity, 63.18%; and cold foot, 59.83%. When symptoms were classified, 21 symptoms were neuropsychiatric (51.2%), 6 were gastrointestinal (14.6%), 4 were musculoskeletal (9.7%), 4 were dermatologic (9.7%), 3 were cardiopulmonary (7.3%) and 3 were unclassified (7.3%).

Group-based findings

The distribution of the patients in the two groups according to the presence of anemia was 168 (70.3%) with IDA and 71 (29.7%) with NAID (Table 1).

Table 1.

Distribution of basic parameters of patients according to iron deficiency anemia and non-anemic iron deficiency groups

	Group	N	Mdn	IQR	Min	Max	Shapiro-Wilk		Mann-Whitney U
							W	P	Statistic	P
Age	NAID	71	39	18.5	21	66	0.960	0.024	5831	0.785
(years)	IDA	168	39	14.0	20	65	0.975	0.004
HGB	NAID	71	12.6	1.0	12.0	14.8	0.900	< 0.001	0	< 0.001
(g/dL)	IDA	168	9.9	2.63	5.2	11.9	0.940	< 0.001
TSAT	NAID	71	16.5	9.6	5.54	55.1	0.875	< 0.001	994	< 0.001
(%)	IDA	153	5.28	4.54	1.91	32.6	0.778	< 0.001
FERR	NAID	69	14.6	9.06	3.32	28.9	0.974	0.166	1488	< 0.001
(mcg/L)	IDA	162	5.04	5.32	0.5	28.9	0.813	< 0.001

Abbreviations: Mdn, median; IQR, interquartile range; Min, minimum; Max, maximum; NAID, non-anemic iron deficiency; IDA, iron deficiency anemia; HGB, hemoglobin; TSAT, transferrin saturation; FERR, ferritin

Statistically significant P values for the Mann-Whitney U test are shown in bold

The demographic and clinical characteristics of the patients are encapsulated in Table 1. The Shapiro-Wilk test indicated that age, hemoglobin (HGB), and transferrin saturation (TSAT) were not normally distributed in both groups (all P <.05). Consequently, these variables are displayed as median (interquartile range). Likewise, serum ferritin (FERR) in the IDA group exhibited a non-normal distribution (P <.001) and is presented as median (interquartile range) in Table 1. In the NAID group, FERR exhibited a normal distribution (P =.166), with a mean FERR of 14.93 ± 6.182 mcg/L.

Groups were analyzed using the Mann-Whitney U test for non-normally distributed variables and the independent samples t-test for normally distributed variables. Table 1 indicates no significant age differences between the NAID and IDA groups (U = 5831, P =.785). Statistically significant differences were noted for HGB, TSAT, and FERR, with the IDA group exhibiting lower median values than the NAID group (all P <.001).

In another analysis, a group comparison was conducted to examine differences in the number of symptoms between the IDA and NAID groups. The IDA group (M = 17.1, SD = 7.22) reported a significantly higher number of symptoms compared to the NAID group (M = 15.0, SD = 6.39), t(237) = − 2.13, p =.034. Prior to the t-test, assumptions were checked. The Shapiro–Wilk test indicated that the distribution of the number of symptoms did not significantly deviate from normality, W = 0.989, p =.070. Levene’s test confirmed homogeneity of variances, F(1, 237) = 0.994, p =.320. Therefore, the assumptions for the independent samples t-test were met.

The distribution of symptoms according to iron deficiency groups is shown in (Table S1) and Fig. 2. The chi-square test was used to compare the proportions of symptoms within the groups. Among the forty-one different symptoms, 35 were proportionally more common in IDA, and 6 were more common in NAID. However, among all symptoms, only palpitations (χ²(1, N = 239) = 7.31, P =.006), cold intolerance (χ²(1, N = 239) = 6.02, P =.014), easy fatigability (χ²(1, N = 239) = 4.08, P =.04) and the “others” symptom class (χ²(1, N = 239) = 4.96, P =.026) were significantly different.

Fig. 2.

Horizontal bar plot graph of the distribution of symptoms according to the groups. The symptom frequencies of both groups exhibited comparable values, resulting in a mirrored or symmetrical representation in the graph. Abbreviations: NAID, non-anemic iron deficiency; IDA, iron deficiency anemia

Associations between symptoms and biochemical parameters in the entire sample

Initially, the analysis of the relationship between symptom diversity and continuous variables revealed no significant correlation with age; however, it demonstrated a significant correlation with hemoglobin (HGB) (r=-.132, n = 239, P =.04) and ferritin (r=-.134, n = 231, P =.04), while transferrin saturation (TSAT) exhibited a more stated correlation (r=-.229, n = 224, P <.001).

The correlation between symptoms and continuous variables was analyzed separately. A correlation test was used to examine the relationships between several variables and symptoms (Table S2).

Upon analyzing the association between symptoms and age, certain symptoms exhibited a significant correlation. Among these, anorexia (r=-.238, P <.001), weight loss (r=-.191, P =.003), diarrhea (r=-.139, P =.03), hair loss (r=-.140, P =.03), onycholysis (r=-.160, P =.01), attention deficit (r=-.179, P =.005), concentration deficit (r=-0. 230, P <.001), cold hands (r=-.228, P <.001), cold feet (r=-.167, P =.01), geosminophilia (r=-.139, P =.03), joint pain (r =.142, P =.028) and hot flashes (r =.135, P =.04) were significantly different.

The correlation of hemoglobin with symptoms included weakness (r=-.142, P =.028), fatigue (r=-.145, P =.024), easy fatigue (r=-.162, P =.012), palpitations (r=-.204, P =.002), and other symptoms (r =.167, P =.01). Notably, an examination of ferritin revealed that only palpitations (r=-.195, P =.003), irritability (r=-.135, P =.04), and “others” (r =.168, P =.01) exhibited a significant correlation with symptoms.

Among the blood parameters, TSAT was most strongly correlated with symptoms, including weakness (r=-.164, P =.014), fatigue (r=-.139, P =.04), and palpitations (r=-.215, P =.0011), onycholysis (r=-.137, P =.04), unwillingness (r=-.141, P =.034), unhappiness (r=-.163, P =.014), irritability (r=-.146, P =.03), feeling cold (r=-.135, P =.044) and pica (r=-.168, P =.012).

Following the correlation test and the observation that TSAT, a blood parameter, was more strongly correlated with symptoms, TSAT median values were compared according to the presence of symptoms. Symptoms that were significantly correlated with TSAT were also significant according to the Mann‒Whitney U test. Accordingly, the median values in the absence of the relevant symptoms (Mdn_a) and the median values in the presence of the symptom (Mdn_b) were significantly different. These were:

Fatigue (U = 2102, P =.015, Mdn_a=10.0%, Mdn_b=6.92%, r =.278), easy fatigability (U = 3509, P =.037, Mdn_a=9.75%, Mdn_b=7.06%, r =.193), palpitations (U = 4628, P=. 0013, Mdn_a=9.64%, Mdn_b=6.45%, r =.250), onycholysis (U = 5075, P =.04, Mdn_a=8.58%, Mdn_b=6.58%, r =.161), unwillingness (U = 5172, P =.035, Mdn_a=8.94%, Mdn_b=6.78%, r =.164), unhappiness (U = 5084, P =.015, Mdn_a=9.28%, Mdn_b=6.77%, r =.188), nervosity (U = 4777, P =.03, Mdn_a=9.68%, Mdn_b=6. 92%, r =.175), feeling cold (U = 4195, P =.045, Mdn_a=8.75%, Mdn_b=7.23%, r =.173) and pica (U = 1194, P =.012, Mdn_a=7.81%, Mdn_b=4.97%, r =.356). Although not significant, the median TSAT values were greater in patients with constipation, neuropathic complaints and “other symptoms”. A graph of the median TSAT values according to the presence of symptoms is presented in Fig. 3.

Fig. 3.

Line graph of TSAT values according to the presence of symptoms. All complaints, with the exception of constipation, neuropathic complaints, and “other symptoms,” were positive under TSAT < 8%. Abbreviations: TSAT, transferrin saturation

Discussion

In this study, forty-one separate symptoms or symptom groups were identified and evaluated. In addition to symptoms with a well-known association with iron deficiency, other less well-recognized complaints were also analyzed. Fifteen symptoms had a frequency of 50% or more in the entire sample. In fact, the average number of symptoms per person was 16.5 (7.03), which shows how patients with iron deficiency have a wide range of different symptoms.

Twelve different symptoms were significantly correlated with age. While 10 of these symptoms were negatively correlated with age, hot flashes and joint pain were positively correlated. The increasing frequency of these two symptoms with age is known [12, 13]. Younger individuals tend to report higher levels of symptoms or complaints than older individuals do, which explains the negative correlation between age and symptoms in this study [14].

Our study found several novel and clinically significant findings. Among non-anemic iron deficiency (NAID) patients, neuropsychiatric symptoms were extremely common. This is crucial because these symptoms are often subtle, easily overlooked, or misattributed in clinical practice, despite their potential to significantly impact patients’ quality of life. The prevalence of such symptoms in NAID suggests that iron deficiency may affect neurological and psychological well-being even without anemia, requiring greater clinical awareness and screening.

In our clinic, weakness, fatigue, and easy fatigability symptoms are handled separately, although they are similar to each other. Among these symptoms, which are also different in definition, weakness was found with the highest frequency [15]. These three symptoms were present in eight out of ten patients. On the other hand, the relationship between fatigue and iron deficiency has been discussed more than that between weakness and iron deficiency in the literature. One study reported a 100% frequency of fatigue symptoms related to these symptoms, especially due to iron deficiency [16]. Although some studies have failed to show an association between nonspecific symptoms such as fatigue and NAID, a meta-analysis confirmed this association [17].

Twelve of the fifteen symptoms with a frequency of more than 50% were neuropsychiatric complaints. This reflects how closely the brain and nervous system are related to iron [18]. These common neuropsychiatric complaints, with the exception of cold intolerance, did not differ significantly in terms of NAID or IDA. In a study that focused on frequency and prevalence, iron deficiency was found in 82.5% of patients with mood and behavioral disorders, 80% of patients with anxiety disorders and 66% of patients with sleep problems [19]. The rate of sleep problems was almost the same as that reported in this study.

Although all the patients in this study were symptomatic and the median symptom variety was more than fifteen, iron deficiency may be commonly symptomatic, it should be noted that there are also iron deficiency patients without any symptoms [20]. However, it should be kept in mind that symptom perception in female patients may be greater than that in male patients [21].

Geosminophilia, a symptom described for the first time in our clinic, was found at higher rates than pica, and further studies specific to this symptom are clearly needed [22]. Although pruritus was observed at a low rate of 2% in this study, the rate of iron deficiency in patients with chronic pruritus was found to be 29% in one study, and the ferritin cutoff in the related study was 15 mcg/L [23].

Another important detail was the rate (18%) of complaints of frequent illness and/or frequent infections in our study. Studies on the association of frequent illness and/or frequent infections with iron deficiency have been conducted mostly in infancy and childhood [24–27]. However, iron deficiency has been reported in 20.1% of patients with recurrent aphthous stomatitis, 3% of those with recurrent infections, and 6% of those with recurrent oral ulcers [28–30]. Notably, few studies have investigated iron deficiency and recurrent disease/infection in adults.

In fact, the most prominent finding of this study is that anemia is not necessary for almost all of the symptoms that can be observed in individuals with iron deficiency. The distribution of symptoms in the NAID and IDA groups was almost completely symmetrical, which could easily explain the lack of significant differences between the two groups. However, although the symptoms of easy fatigability (82.4% vs. 70.1%), cold intolerance (73.8% vs. 57.5%) and palpitations (50% vs. 31%) were significantly different between the two groups in favor of IDA, the presence of these symptoms in the NAID group should not be overlooked. In another study, 19 different symptoms were examined and the NAID and IDA groups did not differ significantly in terms of symptom frequency [31].

Our study also suggests that transferrin saturation (TSAT) may be a useful and accessible biomarker. A strong correlation existed between TSAT levels and reported symptoms. Beyond its traditional role in assessing iron stores, TSAT may be an early indicator of iron deficiency-related symptoms, even in people with normal hemoglobin levels. A readily available biomarker could help diagnose and treat these symptoms earlier, improving patient outcomes.

Transferrin saturation was correlated with more symptoms than was hemoglobin. This correlation was also stronger with TSAT in terms of symptom diversity. On the other hand, patients who reported the presence of most symptoms or groups of symptoms (38 out of 41) had median TSAT values < 8% for all patients, and nine of these symptoms were significantly different. This contribution of TSAT and thus serum iron to the presence of symptoms, which is greater than that of ferritin, seems intriguing.

This necessitates that physicians who evaluate patients with such complaints refrain from becoming fixated on hemoglobin in the symptom-disease relationship. This study has demonstrated that non-anemic iron deficiency can result in a wide range of symptoms and that the frequency of symptoms can be high. Consequently, it is evident that physicians should prioritize iron parameters, particularly TSAT values, in terms of symptom presence.

The most unexpected finding in this study was that the median TSAT value was greater in patients with constipation or neuropathic complaints than in those without constipation, although this finding was not statistically significant. It is possible that patients with both symptoms also have vitamin deficiencies, and the close relationship between these two symptoms and vitamin deficiencies is well known [32, 33].

This study possesses a number of limitations that necessitate careful consideration for the interpretation and generalizability of its results. The cross-sectional design inherently restricts the capacity to establish definitive causality between iron deficiency and the observed symptoms, as it captures associations at a singular moment in time. Future longitudinal or interventional studies are essential to further explore the causal relationships indicated by our findings. The limited sample size of 239 patients, combined with recruitment from a single hospital and an exclusive focus on women, may restrict the generalizability of our findings to more diverse populations and potentially introduce selection bias. Moreover, the absence of comprehensive data on additional micronutrients (e.g., B12, folate, vitamin D), pharmacological interventions, or other medical and psychological conditions hindered our ability to thoroughly account for all potential confounding variables that could also induce similar symptoms. This limits a more accurate delineation of the intrinsic effect of iron deficiency. The subjective nature of symptom reporting and the absence of a symptom severity scale or Likert-type responses restrict the nuanced interpretation and clinical significance evaluation of the observed symptoms. Future research should focus on integrating more detailed and objective symptom assessment instruments. The extensive array of symptoms documented constrained the thorough examination of each, resulting in only a select few being emphasized in detail. Future research may concentrate on particular symptom clusters through more focused analyses.

Conclusion

There are many nonspecific symptoms that may be associated with iron deficiency, and many of these symptoms are very common. Almost every symptom observed in iron deficiency anemia is also observed at high rates in non-anemic individuals with iron deficiency. Therefore, anemia does not have to be added to the picture for symptoms to occur. Transferrin saturation seems to be a biomarker that is more strongly correlated with symptoms. In addition, comparisons of iron deficiency groups with healthy individuals in terms of symptoms will provide cutoff values in terms of biomarkers of related symptoms.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Abbreviations

FERR

Ferritin

HGB

Hemoglobin

ID

Iron deficiency

IDA

Iron deficiency anemia

Mdn

Median

NAID

Non anemic iron deficiency

TSAT

Transferrin saturation

Author contributions

All authors have accepted responsibility for the entire content of this manuscript and consented to its submission to the journal. The hypothesis was formulated by MO. Both authors were involved in data collection. YK was involved in the ethical application. The first version of the manuscript was written by MO. The final version of the manuscript was reviewed by both authors.

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

Data availability

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

Declarations

Ethics approval and consent to participate

This study was conducted in accordance with the principles outlined in the Declaration of Helsinki. The protocol for this research project was approved by the Ethics Committee of Ordu University with approval number 2022/191. Ethical considerations were duly considered during the research’s planning and implementation process, ensuring that the study’s design and procedures conformed to pertinent standards. Since this study is retrospective and data were collected from medical records without direct patient involvement, obtaining informed consent was not required.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.