Preoperative Anaemia and Associated Postoperative Outcomes in Noncardiac Surgery Patients in Central Region of Ghana

Introduction Several studies suggest that preoperative anaemia (PA) is associated with adverse postoperative outcomes, but little is known about these outcomes in the Central Region of Ghana. This study aims to determine the prevalence of PA among noncardiac surgical patients and its implications for their postoperative outcomes. Methods This study was designed as an observational study; data including demographics and clinical and laboratory results were collected from the patients' records and through interviews. Results A total of 893 inpatient surgical cases undergoing elective and emergency operations, aged 15 years and above with mean age of 44.2 ± 17.0 yrs, were enrolled. The prevalence of PA was 54.3%, mostly microcytic with or without hypochromia (57.2%). The prevalence was higher in females than males (p ≤ 0.001). Preoperative anaemia was significantly associated with prolonged length of hospital stay (OR: 2.12 (95% CI: 1.49–3.10)). Allogeneic blood transfusion significantly prolonged the length of hospital stay (OR 4.48 (95% CI: 2.67–7.51)). 15.5% of the anaemic patients received oral iron supplements compared to 2.2% of nonanaemic patients (p ≤ 0.001). Conclusion Preoperative anaemia is common among noncardiac surgical patients. It is independently and significantly associated with prolonged hospital stay leading to the use of increased healthcare resources. It is also the main predictor for perioperative allogeneic blood transfusions and the use of haematinics.


Introduction
Anaemia is an important public health concern that has negative impact on an individual's health as well as the economic potential of the population [1]. e prevalence of anaemia depends on age, sex, and associated comorbidities such as diabetes, hypertension, and in ammatory conditions [2].
Preoperative anaemia is not an uncommon nding, and depending on the cohort being investigated, the prevalence can be as high as 75% [3]. Preoperative anaemia increases the risk of oxygen depletion, thereby increasing the risk of unfavourable postoperative outcomes [3]. ese patients are either given large amounts of red blood cell transfusions depending on the severity of the anaemia [4] or in nonurgent cases necessitating the postponement of surgery [5] till the haemoglobin concentration stabilises or normalises.
Allogeneic blood transfusion is independently associated with increased risk of infection and other adverse postoperative outcomes, especially in countries with low or medium human development index [6,7]. Furthermore, the treatment of preoperative anaemia with RBC transfusions is more costly as compared to correcting anaemia with haematinics like iron, vitamin B 12 , folate, or erythropoietin substitutes preoperatively [8]. Blood transfusion therapy may also increase the risk of transfusion transmissible infections (TTIs) and alloimmunization in a low-to-middle income country such as Ghana. e estimated TTI risk for HIV, HBV, and HCV from a unit of transfused blood in sub-Saharan Africa is 1, 4.3, and 2.5 infections per 1000 units, respectively, based on a mathematical projection [9]. ese three microbiological agents and syphilis are the ones routinely screened in Ghana.
Malaria, a multiorgan systemic disease, can cause and aggravate anaemia, constituting an additional risk in surgical procedures. Studies have demonstrated high rates of morbidity and mortality after surgery in anaemic patients with Plasmodium falciparum associated malarial disease [10]. In addition to anaemia, other confounding factors such as the comorbid conditions can also impede the postoperative progress of the patient [8].
Preoperative haemoglobin levels are checked before surgery in most patients in Ghana. As far as we are aware, no studies have explored the implications of preoperative anaemia on postoperative outcomes in Ghana. is study was therefore carried out to determine the prevalence of preoperative anaemia in unselected noncardiac surgery patients and their outcomes following surgery in the Central Region of Ghana.

Materials and Methods
is study was designed as an observational study and was carried out from June 2015 to July 2016 in two selected hospitals in the Central Region of Ghana. Ethical approval for the study was obtained from the Institutional Review Board of the University of Cape Coast (approval reference: UCC/IRB/3/1). Written permission was also sought from the Management of the Cape Coast Teaching Hospital, Cape Coast, and Saint Luke's Catholic Hospital, Apam. Written informed consent was obtained from individual patients or their guardians. Eligible patients aged 15 and above had a preoperative assessment within 40 days prior to surgery with at least one preoperative haemoglobin (Hb) level to undergo elective or emergency surgery were recruited. Exclusion criteria included planned day-case surgery and obstetric procedures [3,7,8]. Data were collected using the patient's medical record, operation logbooks, anaesthetic records, nurses review notes, and interviews/interaction with surgeons on admission and discharge of the patient. Data were entered into an Excel 2010 sheet anonymously. Patients who were on admission for more than 15 days before surgery, patients who were on admission for more than 30 days after surgery and those with incomplete data such as missingdate of discharge were excluded from the data analysed. Participation did not require any additional visits or assessments. When data collection was not completed before discharge, the participant's medical record was retrieved from the medical records unit. e administration of RBC and haematinics followed the selected hospitals' protocols. A review of the autologous donation records and interview with the blood bank organiser were used to obtain data on autologous donation and the cost of processing a unit of whole blood. Relatives of patients who brought in replacement donors were also interviewed on the cost of presenting a single possible donor.

Data
Collection. Data were collected on patients' characteristics (sex and age), date of admission, date of surgery, date of discharge, preoperative full blood count result, type of anaesthesia (general or regional), type of surgery done, duration of surgery, comorbidity, sickling status, presence of malaria parasites, date of inhospital mortality, complications before discharge, and date of RBC transfusion. e primary outcome measured was preoperative anaemia. Secondary outcomes were postoperative adverse events resulting in complications before discharge, prolongation of hospitalization, the prevalence of perioperative allogeneic blood transfused, and inhospital mortality. Patients were followed up until they were discharged from the hospital. Complete patient con dentiality was maintained.

De nitions.
Preoperative haemoglobin (Hb) levels were the last Hb estimated within 40 days prior to the index operation; majority of the participants 773 (86.6%) had their last Hb estimated within two weeks prior to the index surgery. e World Health Organization (WHO) criteria for anaemia (women 12.0 g/dl and men 13.0 g/dl) and its subclassi cations (nonpregnant women 15 years of age and above: mild: 11.0-11.9 g/dl, moderate: 8.0-10.9 g/dl, severe: lower than 8.0 g/dl; men 15 years of age and above: mild: 11.0-12.9 g/dl, moderate: 8.0-10.9 g/dl, severe: lower than 8.0 g/dl) were used [11]. Morphological examination of peripheral blood lm for anaemia classi cation was not routinely performed at the selected centres. Hence, anaemia was classi ed based on mean corpuscular volume (MCV) and/or mean corpuscular haemoglobin (MCH) into (i) normocytic normochromic types (MCV ≥ 80 but <100 ; MCH ≥ 27 pg): primary aetiology likely to be anaemia of chronic disease, (ii) microcytic (MCV < 80 ) and/or hypochromic types (MCH < 27 pg): possible iron de ciency, and (iii) macrocytic types (MCV ≥ 100 ): possible folate or B 12 de ciency. e surgical procedures (grade of surgery) were classi ed using the BUPA Schedule of Procedures [12].
Postoperative outcomes included inhospital mortality, prolonged length of hospital stay (LOS) after surgery, and the development of postoperative infection or other complications such as cardiogenic shock before discharge and perioperative blood transfusion. Prolonged LOS was categorized as the proportion of patients with hospital stay greater than the 75th percentile (8.0 days), as used by other studies [13,14].

Statistical Analysis.
Descriptive statistical analysis: continuous data were presented as mean, standard deviation (SD), or minimum and maximum values; categorical data were presented as the number and percentage of individuals in each category; and Pearson's chi-square test was used to compare the proportions between two groups.
Binary logistic regression models were performed using adjusted odds ratios (ORadj) to assess the independent e ects of preoperative anaemia on length of hospital stay, postoperative complications, inhospital mortality, and perioperative RBC transfusion. Only factors remaining statistically signi cant (p < 0.05) from the univariate analysis (sex, surgical specialty, and comorbid conditions) and factors that had been theoretically proven to have e ects (i.e., age, grade of surgery, and cancer) were used in the nal model as covariates. Patients with severe and moderate anaemia were analysed together due to small numbers for the severely anaemic.
Analysis was performed with IBM SPSS statistical package version 21.0, USA.

Patient
Characteristics. In all, 953 eligible patients were enrolled in the study. 60 (6.3%) patients were excluded. e excluded patients included patients whose period of admission days was outside those used in the analysis (48), patients with incomplete data (6), and patients with miscellaneous reasons such as high Hb (6). Hence, after data cleaning, 781 (87.5%) patients from a tertiary hospital and 112 (12.5%) from a mission hospital were included in the analysis. e details of the patient's characteristics are shown in Tables 1 and 2. 433 (48.5%) patients had general anaesthesia (GA), 434 (48.6%) had subarachnoid spinal block (SAB), 10 (1.1%) had SAB and GA, and 16 (1.8%) had other anaesthesias such as local in ltration. ere were a total of 624 (69.9%) elective cases out of which 342 (54.8%) were anaemic, and 269 (30.1%) emergency cases out of which 143 (53.2%) were anaemic. e overall median operating time (interquartile) was 76.0 minutes (50.0,111.0 minutes).

Prevalence of Preoperative Anaemia.
Patients with preoperative anaemia accounted for 54.3% of the study population, out of which 209 (23.4%) presented with mild anaemia, 226 (25.3%) with moderate anaemia, and 50 (5.6%) with severe anaemia. e sex distribution, the mean Hb, the severity, and the classi cation of the anaemia as well as surgical specialty are shown in Table 1. Age did not di er among the groups (p � 0.326). e signi cant association of anaemia and comorbidity is shown in Table 2. Hypertension was the most common comorbidity associated with anaemic patients.
When excluded data were included in the analysis, prevalence of preoperative anaemia was 55.2%; it did not skew the anaemia prevalence.

Age and Sex Strati cation of Anaemic Patients.
e anaemia was more predominant among the 15-49 years group than the 50-90 years (p � 0.003). Using this stratication, comorbidity, severity, and classi cation of the anaemia are shown in Table 3.

Postoperative Outcomes.
Evaluation of postoperative outcomes included the length of hospital stay after surgery, postoperative complications before discharge, and inhospital mortality.
(a) Length of hospital stay: preoperative anaemia patients had a signi cantly higher mean length of hospital stay (6.5 days; range: 1-29 days) than nonanaemic patients (4.8 days; range: 1-27 days) (p ≤ 0.001). is probability continued to increase with decreasing Hb concentration in the preoperative anaemia group. e mean in days was 5.9 for mildly anaemic, 6.7 for the moderately anaemic, and 8.2 for the severely anaemic patients. After logistic regression analysis adjusting for potential confounders (statistically signi cant baseline characteristics and those with theoretically proven e ects), preoperative anaemia remained independently and signi cantly associated with increased hospital length stay (Table 4). (b) Complications before discharge: the details of the complications before discharge are shown in Table 5   13.1 ± 0.9 <0.001 9.9 ± 1.5 13.0 ± 0.8

<0.001
Severity of anaemia Anesthesiology Research and Practice association between inhospital mortality with perioperative RBC transfusion after adjusting for signi cant confounding factors as shown in Table 6. For preoperative transfusion, 61/65 (93.8%) of patients were haemo transfused within 2 weeks to the surgery. Additionally, 70/98 (71.4%) of patients received their postoperative transfusion within 2 days after surgery. Some patients (14/65; 21.5%) were transfused before hospital admission. e blood bank recorded two autologous predonations during the study period. However, the patients involved were not transfused; hence, all transfusions were allogeneic.

Cost Implications of Allogeneic RBC Transfusion.
Data available from the blood bank records at the tertiary hospital and interview with patients indicated an approximate cost of 160.0 Ghanaian cedis (approximately 38.0 USD) for obtaining and processing one unit of replacement donation.
3.6. Exclusion. Excluded data did not skew the overall data, when analysed as shown in Table 8.

Postoperative Malaria.
Postoperative malaria parasitaemia data were available for 423 patients. Positive patients who were symptomatic were treated before surgery. e prevalence rate of malaria parasitaemia in the entire study population increased from 2.5% preoperatively to 3.5% postoperatively (p ≤ 0.001). e prevalence rate increased from 3.2% before surgery to 4.2% (p ≤ 0.001) after surgery among the anaemic patients and from 1.8% to 3.4% among the nonanaemic patients (p ≤ 0.001).

Discussion
A high prevalence of preoperative anaemia (approximately 54%) in unselected noncardiac surgical patients was observed Results are shown as odds ratio (95% CI); patients without preoperative anaemia were used as the reference; OR unadj � unadjusted odds ratio; OR adj−1 � odds ratio; LOS � length of hospital stay.  Results are shown as odds ratio (95% CI); no RBC transfusion was used as a reference; OR adj−1 � odds ratio adjusted for severity of anaemia, age, sex, comorbid conditions, surgical specialty, grade of surgery, and cancer; CI � con dence interval; P-LOS � prolonged length of hospital stay; RBC � red blood cell.
in our study, but most had moderate anaemia. Overall, 57% of anaemic admissions had microcytic and/or hypochromic indices, consistent with possible iron de ciency, but normocytic normochromic indices were prevalent in the elderly (65 years and above) together with a signi cantly higher comorbidities, indicating a possible anaemia of chronic disease or in ammation. e preoperative anaemia resulted in prolonged postoperative hospital stay when compared with the Results are presented as percentages; a more than one treatment per patient is possible, that is, some patients received a combination of pre-and/or posthaematinics treatments; others � oral iron and/or folate administered with vitamin B 12 , erythropoietin, or zinc; no patient received haematinics intraoperatively. In this study, about 31% of preoperative anaemia patients received perioperative transfusion with an overall mean of two units per patient. e independent determinants for blood transfusion were the severity of the preoperative anaemia and comorbid condition. Perioperative transfusion was signi cantly associated with poor postoperative outcomes. is study is in agreement with other studies showing high levels of preoperative anaemia among unselected surgical patients [8] and among elective orthopaedic patients [3]. However, the high prevalence rate of 54% despite the preoperative transfusion and haematinics use contradicts previous studies of about 29% [8], 25% [15], and 14% [3]. Approximately, 47% prevalence rate was reported among nonpregnant Ghanaian women of reproductive age (15-49 years) in the Central Region of Ghana using the WHO criteria [16]. But our ndings showed a higher prevalence rate of about 66% among women in the reproductive age group even though women formed 50.8% of the study population. Olayemi and colleagues reported approximately 52% anaemia among apparently healthy young Nigerian adults [17].
Saleh and colleagues [18] reported 23% possible iron deciency among 1142 elective major joint arthroplasty patients aged 15-91 years using hypochromasia as an index. Our study recorded approximately 57% anaemic patients with microcytosis and/or hypochromasia. About 64% of females in the reproductive age group had microcytic and/or hypochromic indices. is is in line with the high prevalence of iron deciency among black Africans including Ghanaians. A prevalence of 32.0% was reported among Ghanaian women living in Germany [19], 41-63% iron de ciency among Ivoirians women [20], and 39% IDA among Nigerians women aged 16-45 years [21]. alassaemia is equally prevalent in Ghanaians, alpha-thalassaemia is about 34% [19] and betathalassaemia is about 5% [22], but could not be di erentiated due to restriction on con rmatory investigations at the selected centres. Anaemia of chronic diseases could not also be excluded.
61% of the elderly patients had anaemia which was mostly mild with normocytic normochromic indices. ey also had signi cant comorbidities suggesting possible anaemia of chronic disease or in ammation.
Regarding outcomes, preoperative anaemia remained an independent risk factor for prolonged hospital stay after adjusting for potential confounders. Some observational studies in noncardiac surgical patients emphasize this association [9,23]. e overall composite complication before discharge was 13.4%. Although comorbidity was signi cantly greater in preoperative anaemia patients, the anaemia did not significantly modify the composite complication. However, other studies have found anaemia to be an independent predictor for postoperative complications [6,8]. is is probably due to the inability of this study to capture complications such as renal insu ciency. e short duration in which the complications were recorded may be a contributing factor.
Our study did not show a signi cant association between preoperative anaemia and inhospital mortality. is agrees with the ndings of Greenky and colleagues [24] and Munoz and colleagues [25]. However, Baron et al. [8] found preoperative anaemia to be signi cantly and independently associated with inhospital mortality. Saager and colleagues [15] found anaemia to be a weak independent predictor of 30-day postoperative mortality.
Our study is also in agreement with previous studies showing preoperative anaemia to be an independent and strong predictor for increased perioperative RBC transfusion [3,7]. e result of this study suggests that allogeneic RBC transfusion is signi cantly associated with prolonged length of hospital stay, higher complications, and higher mortality. After correcting for patients characteristics, prolonged length of hospital stay remained independently and signi cantly associated with allogeneic RBC transfusion [3].
is study recorded a relatively low perioperative transfusion rates compared to the high prevalence of preoperative anaemia. is may be re ective of the type of surgery, the sample size, and amounts of blood loss (which was not captured). Some patients could not get replacement donors and had to be discharged with fewer units of blood transfused or without being transfused when the Hb level was not lifethreatening. In the hospitals where the research was carried out, voluntary blood donation is not common; hence, patients have to rely on replacement donation which can be di cult to obtain. Transfusion attracts an extra nancial burden on the patient which most cannot a ord.
Despite recent recommendations advising investigating the cause of anaemia and treating it preoperatively to minimise transfusion requirements in elective surgical patients, the percentage of anaemic patients treated preoperatively was very low, about 12%. Preoperative oral or parenteral iron or postoperative parenteral iron with or without erythropoietin-stimulating therapy is useful for the management of perioperative anaemia [8,18].
Traumatic events like tissue damage, blood loss, and surgery trigger cascades of in ammatory mediators to cause immunodepression with increased risk to develop malaria parasitaemia [26]. In a study among traumatized land mine and war victims in Cambodia [27], about 33% of asymptomatic malaria carriers developed posttrauma and postsurgery symptomatic malaria. Our study showed a signi cant increased prevalence rate of postoperative malaria parasitaemia in agreement with these ndings [26,27]. Ghana being an endemic area for malaria, the immunosuppression due to surgery predisposes patients to malaria parasitaemia. However, very few of the patients in this study were symptomatic postoperatively. Our study, however, contradicts the study by Takongmo and colleagues [28], which showed that the development of postoperative malaria was not signi cantly associated with the surgical trauma.

Limitations
About 1.5% of the preoperative Hb concentrations were obtained more than 28 days prior to the surgery and may not have represented the actual Hb level at the time of surgery. Approximately 6.8% of the patients were on preoperative haematinics, and 7.3% were transfused prior to the surgery; however, there was no repeat of the Hb levels for a number of the patients after the initial assessment.
Postoperative Hb levels were not available for most patients; hence, their e ects could not be analysed.
Also, anaemia was classi ed based on only mean corpuscular volume (MCV) and mean corpuscular haemoglobin (MCH); increased reticulocyte count due to compensatory response to anaemia and/or the preoperative haematinics administered to some patients can spuriously raise the MCV value masking microcytosis. Con rmatory tests such as ferritin or soluble transferrin receptor measurement for the diagnosis of iron de ciency are not routinely performed at the selected centres. Additionally, data on renal and liver function were not uniformly available for most patients and hence not captured.

Conclusion
Preoperative anaemia is common among noncardiac surgery patients. It is independently and signi cantly associated with prolonged hospital stay leading to increased healthcare resource use. It is also the main predictor for perioperative allogeneic blood transfusions and the use of haematinics.
is study has potential implications for improving the perioperative management of noncardiac surgery patients in the Central Region and Ghana as a whole. It will enhance adequate preparation and comprehensive care to minimise known complications associated with preoperative anaemia. It will also serve as a pilot for a wider national study on postoperative outcomes among noncardiac surgery patients.