Prevalence of iron deficiency and red blood cell transfusions in surgical patients

Abstract Background and Objectives While iron deficiency (ID) is the most common cause of anaemia, little is known about the prevalence and type of ID in preoperative surgical patients. The aims of the present study were to investigate the prevalence and types of ID in a large cohort of surgical patients, and how these are related to perioperative blood use after correction for confounders such as haemoglobin level. Materials and Methods Data were retrospectively extracted from electronic case records of all patients who underwent elective surgery between September 2016 and November 2017 (n = 2711). Iron parameters, haemoglobin and details of perioperative red cell transfusions were collected. Results Of 2711 patients, 618 (22.8%) were iron deficient (= transferrin saturation [TSAT] < 16%) preoperatively, 173 (6.4% of the cohort) had an absolute iron deficiency (AID; TSAT < 16% and ferritin < 30 μg/L) and 445 (16.4%) had functional/mixed ID (TSAT < 16% and ferritin ≥ 30 μg/L). Corrected for Hb level, iron‐deficient patients received significantly more red cell units than patients without ID (p = 0.026). AID was not associated with a significantly higher incidence of transfusions (7.5% of patients transfused; p = 0.12 after correction for Hb) than patients without ID, whereas patients with functional/mixed deficiency did receive significantly more transfusions (6.1%; p = 0.021) as compared to patients without ID (1.7%). Conclusion Preoperative ID, in particular the functional/mixed type, was associated with a higher risk of receiving perioperative red cell transfusions as compared to patients without ID. Adequately treating ID might, therefore, reduce the need for perioperative red cell transfusions.


INTRODUCTION Background
Patient blood management (PBM) refers to the application of evidence-based medical and surgical concepts to optimize the preoperative haemoglobin concentration and haemostatic potential and to minimize blood loss during surgery. PBM aims to improve patient outcome by improving low preoperative haemoglobin levels and reducing perioperative red blood cell transfusion (RBCT) support [1][2][3][4][5]. Correction of iron deficiency (ID)-associated anaemia is one of the most applied measures and has received much attention in recent years. ID is commonly found in patients undergoing surgery and is associated with increased risk not only for receiving an RBCT but also of prolonged hospitalization and postoperative mortality and morbidity [6][7][8][9][10].
ID can either be an absolute ID (AID) due to blood loss or insufficient dietary intake, or functional, as a consequence of chronic inflammation leading to insufficient utilization of the iron stores and decreased uptake by the enterocytes [11]. In the case of functional ID, intravenous iron administration is often needed because of the poor enteric iron uptake.
The preoperative assessment of iron parameters is not standard practice in the Netherlands. In risk groups, in which iron parameters are assessed more often, the use of preoperative iron therapy has become a pragmatic standard of care, with orthopaedic, abdominal and cardiac surgery using the intravenous route of administration as the most effective and fast-acting modality [1,3,[12][13][14][15].

Objectives
Our objectives are to investigate the preoperative prevalence and type of ID and whether the different types of ID are associated with perioperative RBCTs. If such an association is found, the presence and type of ID will be relevant to define the target population in which to evaluate whether iron administration has an impact on perioperative RBTC and the clinical outcome. If the effect depends on the type of ID, this will allow the identification of patients who may benefit from iron supplementation in a costeffective way [7]. Therefore, we performed a retrospective study in a large cohort of surgical patients to investigate whether ID is associated with perioperative RBCTs, corrected for predefined confounders such as haemoglobin (Hb) and whether the transfusion requirement is additionally associated with the absolute or functional/mixed type of ID.

Study design
This is a retrospective single-centre study in the Haga Teaching Hospi-

Study population
The study population was comprised of all adult patients who underwent any form of elective inpatient surgery between September 2016 and November 2017. Patients who underwent non-elective or outpatient surgery or in whom iron parameters were not tested were excluded.

Data collection
In September 2016, testing iron parameters (ferritin, transferrin, transferrin saturation [TSAT] and iron) preoperatively (<30 days before surgery) was introduced as the standard of care, but in November 2017, this practise was stopped due to cost reduction considerations. Iron parameters, Hb level, C-reactive protein (CRP), administration of perioperative RBCTs (30 days before to 30 days after surgery), type of surgery, age and sex of patients were collated. Data were obtained from electronic medical records by the authors.

Definition of anaemia, ID and classifications into subtypes of ID
In accordance with the WHO criteria, anaemia was defined as an Hb concentration of <13 g/dl for adult men and Hb < 12.0 g/dl for adult, nonpregnant women [16]. With the collected data, we attempted to determine the origin of the ID (functional ID; AID and mixed). No universally accepted definitions of absolute, functional and mixed ID exist. Therefore, we chose to divide the cases into groups based on TSAT and ferritin. Reference values were taken from published literature, as indicated.
Patients were considered iron deficient when TSAT was <16% [17,18]. In order to assess whether the type of ID was of influence on the need of RBCTs, we made a subdivision for all ID patients: a patient was considered to have AID when ferritin was <30 μg/L and not AID (FMID: functional/mixed ID) when ferritin was ≥30 μg/L [17,19].
The prevalence of ID and the subtypes were calculated with the aforementioned criteria. We compared demographic data and perioperative RBCTs between patients with ID (AID and FMID) and non-ID (TSAT ≥ 16%) to assess the association between ID and RBCTs.

Statistical analysis
Categorical variables are summarized as frequencies and percentages, and compared with the chi-square test; continuous variables are reported as means and standard deviations, and analysed with a oneway analysis of variance.
We stratified our data by anaemia (binary) to evaluate additional value of iron parameters over Hb. Odds ratios are presented to show the association between preoperative ID and perioperative RBCTs.
The Dunn-Bonferroni correction was used to compensate for multiple hypothesis testing. Ordinal regression modelling was carried out to explore the predictive value of iron parameters for RBCTs. In the ordinal regression model, the variables assessed included Hb level (in this case non-binary), ferritin, TSAT, perioperative RBCTs, age, CRP, sex and type of surgery. All analyses were conducted using SPSS (version 25.0, SPSS Inc., Chicago, IL). A p-value <0.05 (two-sided) was considered statistically significant. who were included in our analyses ( Figure 1). Iron parameters were tested <30 days before surgery (median = 21 days). Baseline characteristics of included patients are shown in Table 1.  11.3 AE 0.8 g/dl) and 11.7 AE 0.6 g/dl in the non-ID group.

Prevalence of ID and anaemia
The prevalence of preoperative ID for the different surgical specialties is shown in Figure 3. As can be expected, in each surgical field, the prevalence of ID is significantly higher in anaemic patients than in non-anaemic patients. The prevalence of AID was highest at 20.3% in gynaecological surgery patients (51/251).
Evaluating the surgical specialties separately, we found that more patients in cardiac surgery (p = 0.003) and orthopaedics (p < 0.001)

ID, anaemia and patients receiving RBCTs
We stratified for the presence and absence of anaemia to see whether the association between ID and RBCTs remained. Of all patients with neither ID nor anaemia, 1.35% (26/1930)

DISCUSSION
In this study, we show that 22.8% of patients have ID preoperatively.
Having ID resulted in a four-fold increase in RBCT in our cohort.
While anaemia is more often present in ID patients, stratification for anaemia shows that RBCTs are only transfused significantly more in ID patients if they are also anaemic. In the non-anaemic group, there is a non-significant trend of increased RBCT in ID patients.
Interestingly, the ordinal regression model indicated that patients with FMID received significantly more RBCTs compared to non-ID, whereas patients with AID did not. This could be a first indication that we may need to specifically target FMID patients in order to costeffectively improve PBM. However, these findings may also be caused by underpowering, as the AID group (n = 173) is smaller than the  Abbreviations: AID, absolute iron deficiency (TSAT < 16% and ferritin < 30 μg/L); FMID, functional and mixed iron deficiency (TSAT < 16% and ferritin ≥ 30 μg/L; non-ID, non-iron deficiency (TSAT ≥ 16%); RBCT, red blood cell transfusion; TSAT, transferrin saturation. a p < 0.05 compared to non-ID; no significant differences were found comparing AID to FMID in this table. between FMID and AID, distinguishing between these types of ID may prove to be useful in the context of PBM [21,22].
A recent observational study found that intravenous (IV) iron preoperatively reduced the risk of perioperative RBCTs [23]. In line with our results, Hubert et al. also found associations in elective cardiac surgery patients between ID and anaemia and between ID and the number of RBCTs (p = 0.03) perioperatively [24]. However, they did not correct the latter analysis for the Hb level, nor was the type of ID taken into consideration. Our findings, in contrast, clearly show a relation between preoperative ID and perioperative RBCTs after correction for the Hb level.
A smaller study of 100 patients undergoing cardiac surgery showed results consistent with ours: patients with ID but without anaemia received more RBCTs than patients with neither ID nor anaemia [25].
Conflicting with the latter and our findings, Fotland et al. did not find such a correlation in a group of 175 orthopaedic patients [26].
A small randomized controlled trial in cardiac surgery did not find a reduction in RBC transfusions after oral and IV iron [12]. However, the iron supplementation started when patients were admitted just before surgery, which may very well be too late for effective iron therapy. Another small trial in the UK with anaemic patients undergoing colorectal cancer surgery found that IV iron had no effect on perioperative blood use compared to oral iron supplementation [14], while a small Australian trial found that IV iron for iron-deficient anaemic patients did improve perioperative blood use [3].  [27]. It has been argued, however, that without a standardized approach to transfusion it should not be concluded that improving preoperative red cell mass does not reduce the need for blood transfusion [28].
In contrast to the above-mentioned studies, the present study spanned the whole scope of surgery in a real-life setting, which strengthens the external validity of our results. In general, we found a clear association between ID and the number of RBCTs after correction for predefined confounders, the most obvious being Hb level, but also age and sex. Other possible confounders (e.g., CRP, type of surgery) were also assessed to maximize the chance of finding a true effect.
Our study is limited, first, by its retrospective nature. larger. An additional and more important question to resolve is whether iron supplementation with an increased Hb in consequence also leads to the hoped-for better short-and long-term outcomes.
Although a lower Hb level and a higher amount of blood use are both correlated with inferior outcomes, we should keep in mind that this is likely to be confounded by the severity of the underlying pathology in these patients. Therefore, it remains unclear whether Hb correction by iron supplementation improves outcome. Until this is established, we recommend IV iron over oral iron when aiming to correct ID preoperatively. Particularly, because we found a stronger association of ID with the incidence of transfusions in the FMID patients, in whom the enteric uptake is impaired [11]. In addition, we recommend starting treatment at least 1 week before surgery so that the therapy has sufficient time to exert a beneficial effect.
In conclusion, our data show that preoperative ID, corrected for Hb level, is clearly associated with the number of RBCTs given to patients.
As the association of FMID with the need for RBCTs is stronger than that of AID, our data support distinguishing between the types of ID. Further research should be undertaken to assess the potential and cost-effectiveness of adequate ID correction preoperatively for reducing RBCTs. Treating ID even in patients without anaemia might also contribute to reduced need for perioperative RBCTs.