Our study has revealed a large burden of anaemia among children under the age of 5 years presenting to the MNH-ED. Since laboratory testing was at the discretion of the treating doctor, not all children were screened for anaemia. If we assume the unlikely scenario that anaemia was absent in all of the children who were not screened, we can conservatively estimate that at least one third (257/777) of children presenting to our ED are anaemic. We found that more than two thirds of all children tested in our ED have anaemia as compared to the estimated prevalence of anaemia globally (24.8%) [2], and that of developed countries [15, 16, 23, 24]. However, these findings were very similar to other low resource countries in South and Southeast Asia, and South and West Africa [3, 4, 16, 18]. The prevalence of anaemia in this study was lower compared to the study done in Ghana where more than a quarter of children admitted had anaemia, with as 71% of these children having severe anaemia which is higher compared to our study where children with severe anaemia comprised 42% of children with anaemia [4]. Our findings are also similar to the 2016 Tanzania Demographic and Health Survey (TDHS), which reported the prevalence of anaemia among children under 5 years-old to be 58% [25].
The proportion of anaemia found in our study is slightly lower than that previously reported among the MNH inpatient paediatric population, as documented by Magesa et al. who found a prevalence of 80.7% [8, 9]. Likewise, anaemia was found in 76% of children under 5 presenting to one of several hospitals in East Africa with severe infection [26]. The slightly lower rate of anemia in our cohort is likely due to the fact that we enrolled all children who underwent haemoglobin testing, regardless of their clinical condition or diagnosis. In fact, anemia was much more common in our cohort of children presenting to the ED than the 12% rate of anaemia reported in a study of 53,174 children admitted to one of 10 hospitals in Kenya [12].
The severity of anaemia in our study was striking. Although the proportion of anaemia in our study was similar to what is reported by TDHS [25], the distribution of severity of anaemia was very different. The proportion of anaemic under-fives with moderate to severe anaemia was higher (81.7%, 95% CI: 76.4–86.2%) in our population than the 32% rate of moderate or severe anaemia documented in the TDHS report [25]. This difference may reflect a different burden of disease that presents to MNH-ED. The patient population at MNH is largely referred from other hospitals, including 76.3% of our cohort, and these patients may be more likely to be anaemic and have transfusion indications than patients presenting to regional hospitals, either due to illness severity or simply because they were transferred to MNH specifically for blood transfusion due to lack of blood at the peripheral hospitals [2, 11]. The proportion of severe anaemia also raises the percentage of children meeting WHO transfusion criteria. Furthermore, the proportion of severe anaemia in our study is higher compared to other East African reports (12% in Kenya and 41% in Eastern Uganda) [26]. Despite the proportion of children with severe anaemia being high in our study (42%), it was lower compared to studies in Ghana where up to 71% present with severe anaemia [27]. The causes of anaemia in our study were found to be similar with a report from several Kenyan Hospitals where malaria was the leading cause of anaemia [12].
Our study showed that 48.6% of the anaemic under-fives presented at the ED with clear WHO defined indications for blood transfusion. Despite the high prevalence of anaemic under-fives with indications for transfusion in the MNH-ED, the transfusion rate was low (23.2%). This is quite low when compared to rates from EDs in the developed world [28]. We were unable to find data on transfusion rates in other ED settings in East Africa, but generally among paediatric inpatient populations in East Africa, transfusion rates have also been found to be low (20–45%), demonstrating poor adherence to WHO guidelines for transfusion [10, 26].
In our study, the low blood transfusion rate was associated with a number of different variables. The majority of patients who met transfusion criteria were not transfused because the physician either did not clearly diagnose anaemia or did not identify the indication for transfusion. This finding was in contrast to what was known in the past about physician gestalt. Although physicians in our ED have the ability to accurately diagnose anaemia based on clinical exam as reported by Sawe et al. [29], anaemia was not documented in many of the children in our study. There is a clear need for physicians to take time to properly examine children and look for signs of anaemia even when the primary chief complaint or referral diagnosis is not obviously related to anaemia or the need for blood transfusion. For example, in our setting some patients with complaints such as burn injuries and foreign bodies had severe anaemia with clear indications for blood transfusion. Moreover, when laboratory tests such as hemoglobin are often ordered, documentation of the results and their impact on clinical decision-making should be recorded in the physician documentation.
Physicians did not order blood transfusion in the ED for the majority of patients (67/125, 53.6%) who had indications according to the WHO guidelines. Even when transfusions were ordered in the ED, they were administered in only half of cases (29/58, 50%). This finding was similar to a study done in Kenya, where the majority of patients did not have an order for blood transfusion and 18% of those who did have an order did not receive the blood [12]. The reasons for these findings deserve further study. Possible reasons include both individual provider and system factors. For example, there may be a lack of clinical knowledge amongst providers regarding the criteria for ordering blood. Or, it could be due to the fact that blood and blood products continue to be a scarce resource at MNH and within Tanzania at large, as shown in previous NBTS reports [14]. According to the NBTS, only one third of the blood products that are needed in Tanzania each year are actually collected [14]. The scarcity of blood in Tanzania is not a novel finding. Blood has remained to be scarce in most developing countries and it significantly affect blood transfusion practices in most of these countries [4, 16, 18, 30]. Finally, some ED providers may find it inconvenient to order blood in the MNH-ED as it can increase the length of stay and contribute to overcrowding in the ED. This concern is illustrated in our study as the median time from door to transfusion is 7.8 (IQR 1.9) hours, which creates a backlog of patients in the resuscitation bay. It may have been hard for some physicians to accommodate these patients who needed to wait for blood for a very long time, especially during busy shifts and night shifts. Among those with indications for transfusion who did not receive it in the ED, only 60% were transfused after admission, further underscoring the importance of identifying and treating these children in the ED.
The median time interval to blood transfusion was very long in this study, nearly 8 h, with the blood processing at the laboratory taking the longest time interval. This may be explained by multiple possible factors including patients spending time on a queue to see a doctor, clearing registration/system issues, or in the laboratory for processing and preparation. This time interval is much longer than the guidelines from more mature healthcare systems, suggesting that uncross matched blood should ideally be available within 10 min, and group specific blood should be available within 30 min [31]. The specific cause of this delay in our department was not captured by our study methodology, but may include variables such as overcrowding in the ED, insufficient staff, a lack of blood products in the blood bank, or a lack of a systematic process to ensure availability of timely transfusion in the ED.
In this study, the overall mortality among patients with anaemia was 31/257 (12.1%, 95% CI: 8.6–16.6%), which is similar to prior reports from other locations in sub-Saharan Africa [32, 33]. The mortality rate among children who had indications for transfusion but were not transfused in the ED was almost three times higher than that of those who were transfused while in the ED (29.2% vs 10.3%, p = 0.032). Therefore, our study furthers the concept that early transfusion is associated with decreased mortality as found in other settings [13]. Our findings were similar to the study done by Lackritz et al. in Kenya among inpatient paediatric population, which showed a mortality benefit when blood was given early on the day of admission for those who presented with indications for blood transfusion as compared to those who received blood later [10]. However, the findings in our study did not take into account the other potential confounding factors that might contribute to the mortality difference.
The results from this study were disseminated internally to the practitioners at the MNH-ED and reasons for failure to identify anaemia and order indicated blood transfusions were explored. These included ED overcrowding necessitating a high patient turnover that did not allow time for physicians to wait for haemoglobin level results, impairing their ability to diagnose anaemia and make appropriate treatment decisions in a timely fashion. As a result of these findings, point of care haemoglobin testing was introduced to help physicians more rapidly diagnosis anaemia and initiate treatment. Furthermore, we provided targeted physician education on the clinical and laboratory assessment of anaemia, the WHO transfusion guidelines, and the importance of timely treatment in optimizing patients’ outcomes.
Limitations
Our study limitations include that it was based at a single centre, which serves as the only full capacity ED in the country. This may limit the generalizability of our results. Another limitation is that we relied on physician documentation of anaemia and transfusion indications and this methodology may have underestimated recognition of these clinical diagnoses due to poor documentation. Despite this, our results are striking in that less than half of the patients with indications for transfusion had a transfusion order placed in the ED and only half of those for whom an order was placed were actually transfused. Finally, some paediatric patients presented to the ED during the study time period had no haemoglobin level measurements, which lead to their exclusion from the study; however, it is likely that some of those children were anaemic, resulting in an underestimation of the overall anaemia burden in our study population.