High rate of inappropriate blood transfusions in the management of children with severe anemia in Ugandan hospitals

Severe anaemia (SA), is a common public health problem in resource limited settings, especially in children under 5 years of age. It accounts for 9.7–29% of total paediatric admissions and 8–17% of hospital deaths in sub-Saharan Africa [1‐6]. SA is a clinical condition commonly caused by severe malaria, acute bacterial infections, micronutrient deficiencies, malnutrition and sickle cell anaemia working singly or in combination [3, 7, 8]. Regardless of the cause, appropriate management involves laboratory confirmation of severity of hemoglobin (Hb) level and prompt correction of the severe anemia by blood transfusion [9]. Access to safe and reliable supply of blood is a major challenge in resource limited settings. Therefore, in order to preserve this scarce resource and reduce the risk of transfusion transmitted infections, a conservative approach is adopted in the guidelines recommended by World Health Organisation [9]. In most resource limited settings a threshold Hb of ≤ 6.0 g/dL is used for transfusion [10].

However, adherence to the guidelines is variable [11]. Blood transfusions are often given without laboratory confirmation of Hb level [12] or documented indications [13], and patients with indications for transfusions are not given blood [14]. Most of the studies that examined clinical use of blood amongst children have done so in children with a known pre-transfusion Hb level [4, 6, 15] usually under controlled research settings [11]. However in routine clinical practice the extent of implementation of the transfusion guidelines in the management of children with suspected SA is not well characterised. In addition, the factors and context specific issues associated with implementation of transfusion guidelines were also not well known. We describe the extent of appropriateness of use of blood transfusion in children admitted with a diagnosis of severe anemia in two regional referral hospitals in northern and eastern Uganda.

A total of 2275 children aged 0–5 years were admitted to Lira and Jinja hospitals during the study period. The median age was 1.5 years (IQR 0.75, 3.0). Fever was the most common presentation while malaria test (blood smear or rapid diagnostic test) was the most common investigation done (Table 1). Overall, 574/2275 (25.2%) were assigned a diagnosis of SA. Hb testing was done for 347 (60.5%) of the children with SA diagnosis, of which only 256 (73.8%) had a pre-transfusion hemoglobin (Hb) level (≤ 6 g/dL) (Fig. 1). The prevalence of assigned diagnosis of SA was higher in Lira than Jinja Hospital (33.5% vs. 20.5% respectively) (Table 1). The overall all-cause mortality in the two hospitals was 4.0% (Table 1).

In this study we found a disturbingly high rate of inappropriate use of blood transfusion in the management of anemic children in two referral hospitals in Uganda. More than half (55.5%) of the blood transfusions were given either presumptively (without documented pre-transfusion Hb done) or when a transfusion is not indicated (Hb > 6.0 g/dl). The major issues related to management of SA included unreliable availability of laboratory equipment and reagents to evaluate haemoglobin level and late presentation of SA children to hospital.

Our findings are similar to those of a Kenyan study that retrospectively reviewed transfusion requests from the paediatric ward of a tertiary hospital to the blood bank [12]. The authors found that 52% of the blood transfusions were requested based on clinical judgement without laboratory confirmation of the anemia. This high rate of inappropriate use of blood transfusion in resource limited setting is unacceptable because it leads to wastage of a scarce resource. In addition, giving blood inappropriately not only exposes children to unnecessary risks associated with blood transfusion like transfusion reactions etc. [18] but also deprives patients who are truly anemic opportunity to get life-saving blood transfusion [14].

In the study hospitals, a diagnosis of ‘severe anemia’ was the indication for blood transfusion in almost all (95%) children that were transfused. Conversely, children not diagnosed with SA were unlikely to be transfused, similar to a Tanzanian study where patients not identified by physicians in the emergency room to have severe anemia were not transfused [14]. Thus correct diagnosis of severe anemia is a critical step in the appropriate use of blood transfusion in resource limited settings.

The clinical diagnosis of SA is based on the finding of severe pallor on examination [19]. Compared to laboratory testing, diagnosis based on severe pallor overestimates the burden of SA due to its low specificity in detecting severe anemia [20, 21]. In this study a quarter of the children (25.3%) admitted in the two hospitals were assigned a diagnosis of severe anemia. However amongst SA children with Hb testing done, 26% had Hb > 6.0 g/dL suggesting that about a quarter of the children with clinical diagnosis of SA were not severely anemic and should not have been given the SA diagnosis.

One of the main reasons for the reliance on clinical diagnosis of SA was the lack of reliable equipment to measure Hb whenever needed. Overall Hb was measured in only 60% of suspected SA patients. Confirmation of SA diagnosis by Hb measurement should be available to the clinician as soon as possible to guide clinical decision on whether or not to transfuse. The standard method of measuring level of Hb involves using automated hematology analysers like the coulter counter machine [22]. However, hematology analysers are expensive, requiring ongoing costs for blood tubes and reagents, which is challenging in resource limited settings. They also require greater volumes of blood and a longer turnaround time which is not ideal in emergency conditions like SA where a rapid Hb measurement is required to make immediate management decisions. In this study the Coulter Counter machines in both hospitals frequently broke down because of overload or when they were operational they often lacked reagents. Hence Hbs were often not measured. The alternative would be to use point-of-care haemoglobinometers such as the HemoCue [22, 23] for estimating haemoglobin level. These devices require only a small finger prick for capillary blood sample and provide an estimation of Hb levels in real-time to facilitate objective clinical decision making by front line health workers. Moreover they can easily be used by front line health workers [24] and are fairly accurate if properly used [25]. Although point-of-care hemoglobin tests still require costly consumables like cuvettes, an investment in these point of care devices for measuring Hb would lead to improved inpatient care and help manage limited blood stocks in resource limited settings where mortality from severe anaemia remains unacceptably high.

Poor clinical habits contributed to inappropriate blood transfusions and lack of adherence to transfusion guidelines. In this study laboratory results were disregarded and at least 16% of the patients were transfused inappropriately with pre-transfusion Hb > 6 g/dl. This was lower than the 46% reported in children transfused with Hb > 6 g/dl in the Kenyan study [12]. The main reason given by the clinicians for none-adherence to blood transfusion guidelines was that some of the SA children present late and in critical condition so they prioritised giving blood transfusion without waiting for Hb testing or Hb result. This suggests that children transfused inappropriately were sicker than those deemed stable enough to wait for Hb testing. This is partly supported by the significant difference in inpatient mortality rate between SA children transfused as per guidelines compared to those transfused inappropriately on clinical grounds (p = 0.03). However many of the SA children reported to have come in critical conditions presented during the night shift when the staffing levels are low and composed mostly of junior level cadres. It is therefore more likely that the higher inpatient mortality in the SA children transfused on clinical grounds was because these children were misdiagnosed and inappropriately managed.

Efforts to improve transfusion practices should target clinicians who make the decisions of whether or not to transfuse. Transfusion guidelines also need to be tailored to take into consideration local needs and perceptions amongst health workers and the role of point of care bedside devices for hemoglobin measurements need to be explored. One such training modality is the emergency triage assessment and treatment plus admission care (ETAT+), which has been shown to result in improvement in knowledge [26], clinical skills [27], adherence to clinical guidelines [28] and clinical care [29]. To be effective, it is important for the training to be coupled with interventions that address gaps in health systems such as accessibility of care [30], supply chain issues, staffing levels, [31] remuneration and other such issues that affect staff morale and performance [32].

Delays in getting blood has been reported as one of the issues affecting clinical use of blood in other studies was not an issue in this study [33]. In this study, when blood was available, most of the children were able to receive the blood shortly upon request. This was probably due to the location of the blood banks within or close to the children’s wards. However, as has been the case in settings elsewhere [34, 35], blood stock-out was frequently reported despite the fact that both hospitals were blood collection centres. During blood stock-outs, severely anemic patients were advised to go to other health units where blood might be available. It is ironical that despite this chronic shortage, blood when available was given indiscriminately with disregard to transfusion guidelines. This tendency to transfuse on clinical grounds inflates the demand for blood leading to unnecessary blood transfusions and compounds the shortage of blood. In Uganda, 60% of an estimated 220,000 unit of blood collected annually is consumed by children [36]. If about a half is given inappropriately, as seen in this study, this translates to about 66,000 units of blood per year, more than the projected annual shortfall of 30,000 units annually [36].

Efforts to improve availability of blood for clinical use should include measures to reduce the demand for blood by addressing the causes of severe anemia. In our setting malaria was the most common etiological factor associated with SA. Efforts to control malaria should be strengthened as reduction in malaria rates has been reported to result in corresponding reduction of demand for blood [37].

There were several limitations in this study. Since this was a retrospective study of routine clinical records, it is possible that some of the Hb tests and results were not documented in the inpatient records. Our findings, therefore may underestimate the prevalence of laboratory confirmed SA. Secondly during stock out of blood suspected SA patients are often advised to go to other health units that might have blood. Such missed cases were not recorded so our findings only describe the transfusion practices when blood was available at the health facility. Thirdly, the care was not uniform and depended on the motivation, clinical skills and documentation of the clinician that attended to the patient. Nonetheless, by covering a large number of records over a 12 month period, we believe the data captured in this study presents the average care provided to SA children in Uganda. The major strength of the study is that it provides much needed documentation on blood use during routine clinical practice in resource limited settings.

In conclusion, our study highlights the unacceptably high rate of inappropriate use of blood transfusion in the management of anemic children in Lira and Jinja hospitals in Uganda. Verification of Hb level by laboratory testing and training of health workers to adhere to transfusion guidelines could result in a substantial decrease in inappropriate blood transfusion, which is critical with the chronically low supply of blood in Ugandan hospitals. Future studies should explore the effect of adherence to clinical guidelines on inpatient mortality.