Characterising Childhood Blackwater Fever and Its Clinical Care at Two Tertiary Hospitals in Eastern Uganda

In eastern Uganda, reports suggest that cases of Blackwater Fever (BWF) are on the rise. We summarise the base-line characteristics and routine care available to patients with BWF presenting at two tertiary hospitals in Eastern Uganda prior to the Phase I/II trial on use of paracetamol for acute kidney injury in children with BWF (PARIST; ISRCTN84974248). This was a retrospective descriptive study for the period January – December 2018 for children admitted with a clinical diagnosis of BWF at Mbale and Soroti Regional Referral Hospitals in Eastern Uganda. Data on sociodemographic and clinical characteristics, routine in-patient care and outcomes were abstracted using a customised study proforma and analysed using STATA.


Page 3/19
Background Blackwater fever (BWF) a clinical condition characterized by an acute intravascular haemolysis resulting in passing dark, tea or Coca-Cola coloured urine (1,2), is almost exclusive to P. falciparum malaria (3). Historically, the case de nition included: Caucasian who had lived or visited malaria endemic area for a long time (> 3months) without previous exposure to malaria and were taking quinine in inappropriate dose or schedule for malaria prophylaxis and/or treatment (4)(5)(6). The initial attack of malaria complicated with BWF would typically present with loin pain, abdominal discomfort, restlessness, vomiting, diarrhoea, polyuria followed by oliguria and passage of dark red or black urine. Signs included tender hepatosplenomegaly, profound anaemia and jaundice (1,(7)(8)(9)(10). Since then, however, use of such case de nitions in the African populations have been done with mixed outcomes across geographical or population strata. For instance, prevalence and mortality range from 6-59% (11)(12)(13)(14) and 4.4-25.3% (15)(16)(17) respectively. BWF studies in African children are complex (7,11,18). Adult case de nitions are considered inappropriate for children, where malaria related renal failure was considered a rare clinical feature (19)(20)(21)(22). As a contrast, BWF in children is often at initial stages of illness as opposed to adults who manifest it at the end stages of malaria disease (10). Moreover, there is a low clinical index of suspicion among attending clinicians and limited capacity to investigate the differentials of dark urine in resource-limited settings. To date, there is a paucity of recent descriptions on BWF, which focus on children in Africa (7,11,18). Most referenced data describing the phenomenon of BWF report varying prevalence of 6-48% (11,12) and 11-59% (13,14) of patients with severe malaria respectively. However, most of these were observational studies in adult series involving malaria non-immune patients. Even with a recognised malaria epidemiological transition (23,24), with reports of malaria reducing (25)(26)(27)(28)(29), malaria remains endemic in some regions and countries including Uganda (30). In such settings childhood severe malaria studies suggests a high prevalence of BWF. For instance, 25.3% in Kinshasa in DRC (31), 19.1% in Ibadan, Nigeria (32) and 17.2% in Togo between 2000 and 2002 (16) in contrast to a lower prevalence reported in a large randomised controlled trial of Artesunate v Quinine (AQUAMAT) in African children < 15 years of age with severe malaria (n = 5,246 at 10 sites in nine countries) that indicated BWF in 237/5,426 (4.4%) (33). Earlier studies in Eastern Uganda reported a high frequency of BWF 14.5% and 21.8% at Mbale and Soroti Regional Referral Hospitals (FEAST Trial; ISRCTN69856593) (34). As with AQUAMAT study, the FEAST study was on a highly selected group of patients, which were not representative of the facility-based disease burden. Speculation on the cause of the notable increase in frequency of BWF in this region has been linked to a possible change in the malaria treatment protocols in the last decade (34), but have not been proven. It is also possible that changes to local parasite population to a strain with a greater propensity to cause BWF are in play as hypothesised by other studies (35,36), but this has also not been studied in the region. It is possible that BWF in the region is multifactorial and that is conceivably the reason for high number of cases. In this study, we report BWF in general admissions in Eastern Uganda.

Study Area and Sites
The study was carried out on the admission medical records at the paediatric units at Mbale and Soroti Regional Referral Hospitals (RRH) in Eastern Uganda. Both facilities are government run, not for pro t and charge-free hospitals. Mbale RRH is located at the heart of Mbale City, 214KM to the east of the Capital City Kampala while Soroti RRH is located at the centre of Soroti City; 100KM north-east of Mbale City. At both facilities, four major general specialties: surgery, internal medicine, obstetrics and gynaecology, and paediatrics are offered. Mbale RRH is larger with 470 bed capacity, compared to Soroti RRH with 274 bed, but they have comparable proportions of paediatric bed capacity at 95 (20.2%) vs. 64 (23.4%) respectively. The recent annual paediatric caseload at Mbale RRH on average is 10,000 admissions from a catchment of 16 districts with a population of 4.5 million. The geographical catchment area for Mbale RRH (Elgon sub region) is hilly and situated in the range 980 to 1,800m above sea level. Conversely, Soroti RRH admits approximately 6,000 children annually from a catchment population of 2.9 million in 9 districts. Its catchment geographical area (Teso sub region) is generally at and most districts are < 1000m above sea level.

Study design
This was a retrospective descriptive study. Records for children aged 2 months to 15 years in the paediatric units at the two hospitals with admission diagnosis of BWF for the period January 1, 2018 to December 31, 2018 were studied.

Study Procedures
The paediatrics admission registers were used to identify all the children who presented with a diagnosis of BWF. BWF was diagnosed clinically as passing dark/tea/Coca-Cola coloured urine and corresponding urine colour grade ≥ 5 on the Hillmen Urine Colour Chart (37) (Fig. 1). On admission, the axillary temperature was measured with a digital thermometer. Fever (axillary temperature ≥ 37.50C) was further categorized as mild fever (37.5-37.90C), moderate fever [(38.0-38.50C) and severe fever (> 38.50C). Prostration was de ned as inability to sit upright / stand unsupported /breastfeeding age < 6 months.
Comorbidity was de ned as 'any distinct additional clinical entity that has coexisted or that may occur during the clinical course of a patient who has BWF. Level of consciousness was assessed using the AVPU score ('Alert,' 'Responding to Voice,' 'Responding to Pain only,' or 'Unresponsive'). Clinical jaundice was yellowing of mucous membranes noted in su cient daylight (38). Clinical assessment of pallor of mucous membranes indicated the degree of pallor (either none, mild, moderate or severe). The admitting clinician (Clinical o cer or medical o cer) carried out an assessment of pallor and determined whether they felt the clinical severity of pallor would warrant transfusion.
Trained research assistants retrieved the case records and extracted data on socio-demographic characteristics including gender, age and ethnicity. They further obtained admission data on presenting symptoms, duration of illness, physical signs, co-morbidities and treatment received. Records with missing study variables on age, gender, illegible les, those outside the study period and age range were excluded.

Use of the Hillmen Urine Colour Chart
At admission, patients with a history of passing dark urine during the course of their current illness were initially assessed by clinicians based on history, an inquiry about the child passing dark urine, de ned as Coca-Cola or tea coloured urine on the day of admission was made. Furthermore, asking the parent/guardian to indicate the grade (by pointing) at the colour against the Hillmen Urine Colour Chart (HUCC) (Fig. 1) qualitatively assessed the urine colour. The HUCC had 10 colour codes ranging from mild yellow (colour code 1) to black (colour code 10). A probable and con rmed diagnosis of BWF/dark urine was made at admission. The patient/guardian was rst asked to recall and match the colour of urine passed by their child on the day of admission to a colour on the HUCC. If available, urine was collected from the child using paediatric urine collection bags before it was transferred into the urine collection bottle. The study clinician then matched the urine colour to the corresponding score on the HCC scale. Children with clinician-witnessed urine or patient/guardian matched the colour of urine passed by their child on the day of admission corresponding to HCC > 5 on the chart were con rmed to have BWF/dark urine syndrome. These were the eligible patients for the study. For easy access and e cient use of the HCC, charts were displayed both in the patient admission area and in the ward.

Data management and statistical analyses
The data was entered into MS Excel, exported and analysed using STATA (version 14.0, College Station, Texas 77845 USA). All qualifying records for the period of study chosen were included. Initial descriptive and univariate analyses were carried out on the data. Means and standard deviations (SDs) were determined for normally distributed continuous variables and medians and interquartile ranges (IQRs) for non-normally distributed variables. Proportions and percentages were determined for categorical variables. Study participants were strati ed by age (≤ 5years and > 5 years) and gender. Between group differences were assessed using Pearson's χ2. The difference was considered signi cant at P < 0.05.

Results
During the 12-month study period, 9578 records of children admitted to the Paediatric Acute Care Unit (PACU) at the two hospitals were retrieved. Of these 1308 (13.7%) of the records had BWF. Excluded were 67/1308 (5.1%) records because of the missing data on key variables. The remaining 1241/9578 (13.0%) were eligible records and included in the study.

Pattern of distribution of BWF cases in 2018
Noticeably, presentation with BWF was more in the second half of the year in 2018 with a peak in the month of July to September (Fig. 2). The patterns of distribution of BWF cases disaggregated by sex and age category revealed an overall higher number of BWF cases above 5 years compared to under 5 years for the females while the number of BWF cases under 5 was slightly higher than the above 5 for the males in the two age categories. For both the peak in the number of cases was reported in the second half of the year with the highest being the month of August. (Fig. 3).

Routine treatment received by BWF patients
The routine treatment of patients with BWF is summarized   (31). This has led to questioning the link between BWF and malaria. If malaria was a direct driving factor, then children aged > 5 years who are expected to have acquired malaria immunity (1, 18, 42, 43), would be resilient to BWF. On the other hand, if it was an innate factor then either more of the ≤ 5year old children or a similar proportion across the various age strata would have been observed with BWF. Therefore, greater understanding is still needed to elucidate the aetiopathophysiology of BWF.
A number of speculations have been made to try and explain high prevalence of BWF in these settings. For instance, the coincidence of increased cases of BWF with a roll out of Artemesinin Combination Therapies (ACTs) for control of malaria has postulated but has not been well studied (34). Elsewhere, lumefantrine, a key drug in rst line ACTs has been implicated in the causation of BWF, but descriptions in African children have not been done. Similarly, the role of repeated exposure to other antimarials (18, 31,39), and blood transfusions in the causation of BWF in children have also not been explored in these settings. High rates of index and repeat blood transfusions, in a phenomenon similar to isoimmunisation in mother-baby situation may be underlying in some of the massive haemolysis observed in these populations, but needs further research. Conversely, whereas past epidemiological studies have indicated that the interaction between host response to repeated malarial attacks, use of antimalarials, and possibly glucose-6-phosphate dehydrogenase (G6PD) de ciency are trigger factors (50), more recent descriptions in Eastern Uganda do not associate the phenomenon to G6PD de ciency (34). Alternatively, other scholars have argued that the possible cause is likely a recent change in the malaria parasite population towards a strain with a greater propensity towards causing BWF (35,36).
The seasonal trend of BWF admissions in 2018 showed one peak in the months of July to September. Eastern Uganda receives moderate rainfall with annual rainfall averages ranging from 1100-1200mm and this is distributed between two seasons (March to July and September to November) (51). A malaria surveillance study done in all the regions of Uganda between 2015 to 2019 showed the highest peaks in monthly trends in regional malaria incidence rates were in June-July, highest in June, 2017 (Range: 13.4-95.6 cases per 1000) and July, 2019 and the lowest troughs in February-March of each calendar year (52). Teso region home to one of the study sites (Soroti RRH) was one of the regions that persistently recorded the highest monthly incidence rates across the entire study duration (52). Possibly, the malaria incidence rate peaks of June -July of 2018 may have coincided with the peak in BWF admissions reported in this study between July and September suggesting either similar risk factors or a cause-effect relationship between BWF and malaria in this setting.
Understanding paediatric BWF in Africa is very important for informing treatment options and plans. Currently, there are no speci c treatment and prevention remedies for BWF. According to the Uganda clinical guidelines, all cases are treated as severe malaria and respective supportive treatment such as blood transfusion is given (53), which is plausible. However, with no speci c management guidelines for BWF, inconsistences exist in patient care owing to the differences in clinician expertise/training, experience and acumen. This explains the variability in treatment reported in this study.

Availability of data & materials
The study data are available by request to the corresponding author.

Competing interest
The authors declare no competing interest.  Seasonal variation of BWF cases by sex and age category in 2018.