Quality of care for adult in-patients with malaria in a tertiary hospital in Uganda

Prompt detection and appropriate treatment of malaria prevents severe disease and death. The quality of care for adult malaria in-patients is not well documented in sub-Saharan Africa, particularly in Uganda. The study sought to describe the patterns of malaria diagnosis and treatment among adult in-patients admitted to the medical and gynaecological wards of Uganda’s 1790-bed Mulago National Referral Hospital from December 2013 to April 2014. A prospective cohort of 762 consented in-patients aged ≥ 18 years was assembled. Proportions of in-patients who received preadmission and in-hospital anti-malarials, missed Day 1 dosing of hospital-initiated anti-malarials and/or had malaria microscopy done were determined. Multivariable logistic regression was used to identify risk-factors for missed Day 1 dosing of anti-malarials. One in five (19%, 146/762) in-patients had an admission or discharge malaria diagnosis or both; with median age of 29 years (IQR, 22–42 years). Microscopy was requested in 77% (108/141) of in-patients with an admission malaria diagnosis; results were available for 46% (50/108), of whom 42% (21/50) tested positive for Plasmodium falciparum malaria parasitaemia. Only 13% (11/83) of in-patients who received in-hospital injectable artesunate (AS) or quinine (Q) received follow-up oral artemether-lumefantrine (AL); 2 of 18 severe malaria cases received follow-up oral AL. Injectable AS only (47%, 47/100) was the most frequent hospital-initiated anti-malarial treatment followed by injectable Q only (23%, 23/100) amongst in-patients who received in-hospital anti-malarials. A quarter (25%, 25/100; 95% CI: 17–35%) of in-patients missed Day 1 dosing of hospital-initiated anti-malarials. Each additional admission diagnosis was more than two-fold likely to increase the odds of missed Day 1 dosing of in-hospital anti-malarials (aOR = 2.6, 95% CI: 1.52–4.56; P-value = 0.001). Half the malaria microscopy results were not available; yet, the rate of testing was high. The majority of in-patients initiated on injectable AS or Q did not receive the recommended follow-up oral AL. One in four in-patients delayed to initiate hospital anti-malarials by at least one calendar day. The hospital should encourage prompt availability of malaria test-results to promote the timely initiation and completion of anti-malarial treatment, thereby improving the quality of care for hospitalized malaria patients in Uganda.

is highest during the first 24 h of hospitalization [2]. Yet, in most moderate-to high-malaria-transmission settings, long transit-time to a suitable health facility where appropriate intravenous anti-malarials can be administered could delay the initiation of appropriate anti-malarials and increase the risk of patient deterioration or death [2]. Other impediments to the timely initiation of appropriate anti-malarials include the lack of timely laboratory diagnosis and drug stock-outs [2,3].
The WHO recommends confirmation of malaria diagnosis by quality microscopy or malaria rapid diagnostic testing within 2 h of patient presentation and before administration of anti-malarials. Otherwise, the decision to treat should be taken on clinical grounds. If severe malaria is suspected, parasitological diagnosis should not delay initiation of anti-malarials [2]. Adults with severe malaria, including pregnant women in all trimesters and breast-feeding mothers, should be treated with three doses of injectable artesunate (AS) for 24 h minimum at 0, 12 and 24 h regardless of whether the patient can tolerate oral treatment earlier. If unable to take oral medication, the patient should continue with injectable AS once daily, for a maximum of 7-days. If injectable AS is not available, once daily injectable artemether (AT) or 8-hourly injectable quinine (Q) should be administered. Following injectable anti-malarials, a full 3-day course of oral artemisinin-based combination therapy (ACT)mainly artemether-lumefantrine (AL) (six doses) for Uganda-should be administered if the patient is able to take oral medication [2,[4][5][6]. Other recommended artemisinin-based combinations include artesunate-amodiaquine (AQ) and dihydroartemisinin-piperaquine (DP) (three doses for each combination). If full treatment for severe malaria is not possible at a given health facility but injectables are available, adults and children should be given one intramuscular dose of AS or Q and referred to a suitable facility for appropriate management [2].
It is necessary for patients with severe malaria to initiate timely appropriate anti-malarials and complete full courses of prescribed anti-malarials, which promotes therapeutic success, reduces malaria-related mortality and prevents drug resistance [7][8][9]. However, SSA patients with severe malaria frequently receive incomplete doses of prescribed anti-malarials and/or treatment meant for uncomplicated malaria [8,10]. One-third of hospitalized Ugandan patients missed Day 1 of prescribed antibiotics [3], but similar data are scarce on antimalarial use. This study aims to describe the patterns of malaria diagnosis and treatment [i.e. anti-malarial use by extent of use, missed opportunity for treatment, frequency of administered-treatment, medication-usecycle (prescription-dispensing-administration), missed Day 1 dosing and mortality] among adult in-patients at Uganda's 1790-bed Mulago National Referral Hospital. The study also evaluates patient-level risk-factors for missed Day 1 dosing of prescribed anti-malarials and the relationship between missed Day 1 dosing of prescribed anti-malarials and length of hospital stay among these Ugandan adult in-patients.

Methods
Detailed account of the study design, setting, data collection and data management is documented elsewhere [3]. Briefly, key details are described below.

Study design and setting
A prospective cohort study was conducted among inpatients, 18 years and older, at Mulago National Referral Hospital with bed capacity of 1790 and an annual in-patient turnover of 140,000 patients. Three medical wards and one Gynaecological ward were included, each with an official bed capacity of 54 and average of 5-25 patient admissions per day [3]. Not all women of childbearing age have a pregnancy test done.
Details on the prescription, dispensing and administration of medicines are hand-recorded in the patients' charts. Hospital pharmacists dispense the prescribed free-of-charge injectable and/or oral anti-malarials, as appropriate, to in-patients/caregivers in quantities that discourage misuse of medicines. However, the inpatients/caregivers are instructed to return to the pharmacy early enough for more medication to avoid missing treatment [3].

Data collection
The data were collected in December 2013 to April 2014 by four teams of trained research assistants. Each research team had a medical officer, clinical pharmacist and degree nurse. An internist on the medical wards and a gynaecologist/obstetrician on the gynaecology ward solved any clinical problems faced by the research assistants, while the principal investigator advised on pharmacological issues. The research team did not interrupt routine clinical care. Study patients provided written informed consent and were enrolled at the four wards using a systematic sampling procedure following a daily random start from the first two (Infectious Diseases and Gastrointestinal Illnesses ward), three (Haematology, Neurology and Endocrinology ward) and four (Cardiovascular, Pulmonology and Nephrology ward & Gynaecology ward) new admissions; and subsequently every second, third and fourth admission, respectively. Patients were assessed at baseline (demographics, clinical conditions, medications) and on a daily basis (clinical conditions, medications) until discharge, transfer, death, or loss to follow-up [3].
Research assistants captured clinical data from patients' clinical charts and interviewed the patients, their caregivers and/or ward staff. Each research team's medical officer clerked the in-patients to obtain additional clinical data not documented in the clinical charts. Each research team's pharmacist captured baseline medication data at the time of hospital admission by interviewing the inpatients and reviewing all available medical documents. During hospital stay, research pharmacists obtained medication data from the patients' clinical charts, ward pharmacy records, pill counts of patients' oral medications (tablets, capsules), views of unused parenteral medicines possessed by the patients/caretakers and daily interviews with the patients/caregivers and/or ward staff. The data were collected daily from 8.00am to 6.00 pm from Monday to Friday and from10.00am to 6.00 pm on weekends and public holidays [3].

Data management
Epidata 3.1 software was programmed with checks to limit data entry errors and the electronic database password-secured to limit access to authorized personnel only [3].

Patterns of malaria diagnosis and anti-malarial use
The proportions of in-patients who received anti-malarials preadmission and during hospitalization were determined using, as numerator, the number of in-patients who received at least one anti-malarial and, as denominator, the total number of study in-patients. We calculated the proportions of in-patients with available (positive, negative) and unavailable (not returned, not requested) malaria microscopy results and those who experienced missed Day 1 dosing of hospital-initiated anti-malarials. See Additional file 1 for details of the analysis plan for post-admission time-to-first-dose among in-patients with an admission malaria diagnosis; and parenteral-tooral-switch of anti-malarials.
Chi-squared tests were used to screen univariatelevel relationships between patient-level characteristics and anti-malarial use during hospitalization (yes/no); and potential patient-level risk-factors for missed Day 1 anti-malarial dosing during hospitalization. Multivariable logistic regression was used to identify risk-factors for missed Day 1 dosing of anti-malarials. Results were expressed as odds ratios (ORs) with their 95% confidence intervals (CIs). Poisson CIs were used for counts below 16. Stata 14.0 was used for all the analyses [11].

Identification of missed Day 1 dosing of anti-malarials
Among in-patients for whom an anti-malarial was prescribed during the current hospitalization and at least one dose was administered, missed Day 1 dosing was measured in two ways; (i) calendar-day as proposed by Kiguba et al. [3] and (ii) 24-h timescale using date-andtime of hospital admission and date-and-time of first inhospital anti-malarial dose.

Missed opportunity for hospital-initiated anti-malarials
Four of 25 (16%, 95% CI: 5-36%) in-patients with a positive malaria test did not receive in-hospital antimalarials, see Fig. 1. No admission/discharge malaria diagnosis was recorded for three of the four inpatients, see Fig. 1, two of whom had a malignancy; the fourth in-patient had poorly treated malaria on admission and run away from hospital due to delayed investigations and treatment, see Table S1. None of the in-patients died while in hospital, see Box 1; none had malaria-in-pregnancy.

Overview of prescription, dispensing and administration of anti-malarials
Overall: Anti-malarials were prescribed for 15% (114/762) of in-patients, dispensed to 79% (90/114), yet, administered in 100 in-patients (93 of 100 had an antimalarial prescription), see Additional file 1 for details on AS, Q and AL.
Artesunate or Quinine + Artemether-Lumefantrine: About 13% (11/83) of the in-patients who received inhospital injectable AS or Q also received at least one dose of follow-up oral AL during the current hospitalization, see Table 3; AL having been co-prescribed for 48 (58%) of the 83 in-patients. AL was co-prescribed for 12 (67%) of the 18 severe malaria cases and administered in 2 cases only during the current hospitalization.

Missed Day 1 dosing of hospital-prescribed anti-malarials
Overall: A quarter (25%, 25/100; 95% CI: 17-35%) of in-patients who received anti-malarials during the current hospitalization missed Day 1 dosing of hospital-initiated anti-malarials based on calendar-day. A similar estimate of missed Day 1 dosing was obtained based on post-admission 24-h-delay, see Appendix.

Mortality among in-patients who received in-hospital anti-malarials
Four of 100 in-patients who received in-hospital antimalarials died during hospitalization. All four inpatients had clinically-diagnosed malaria: microscopy was requested is three in-patients, but results were not available, see Box 2. An unconscious 88-year-old female of unknown HIV-status presented with a single admission diagnosis of severe malaria and pulse rate of 98 beats per minute. She received a pre-referral

Patient-level risk-factors for missed Day 1 dosing of anti-malarials
Number of admission diagnoses was a statistically significant risk-factor for missed Day 1 dosing of hospital-initiated anti-malarials based on calendar-day (aOR = 2.6, 95% CI: 1.52-4.56; P-value = 0.001), see Table 5. Similar results of missed Day 1 risk-factor were obtained based on post-admission 24-h-delay, see Table S3. Malaria-inpregnancy and severity of malaria were not significantly related to missed Day 1 dosing of anti-malarials.

Missed Day 1 dosing of hospital-initiated anti-malarials versus length of hospital stay
No statistically significant association was observed between missed Day 1 dosing of anti-malarials and length of hospital stay (OR 1.1, 95% CI: 0.91-1.27; P-value < 0.396). The mean length of hospital stay for missed Day 1 cases was 4.7 (SD = 1.7) days versus 4.2 (SD = 2.5) days for non-cases.

Discussion
Malaria microscopy was requested in 77% of in-patients with an admission malaria diagnosis, similar to estimates for the public sector (80%) in moderate-to hightransmission countries in sub-Saharan Africa (SSA) [1]. Unfortunately, only half the microscopy results were available to guide appropriate anti-malarial treatment. Thus, despite decent microscopy rates, healthcare professionals still rely on clinical judgement to treat half the suspected malaria cases. Clinical judgement increases the risk of unnecessary anti-malarial treatment and, in turn, depletes anti-malarial stocks for in-patients who truly need them; and increases the incidence of associated adverse drug reactions and drug resistance [2]. Seven in ten in-patients with suspected non-pregnancy-related malaria tested negative for malaria and would therefore not need anti-malarials; compared with only two in ten in-patients with suspected malaria-inpregnancy. The value of a confirmed malaria diagnosis depends on prompt availability of parasitology results and whether the clinician uses the results to decide how to manage the patient. Malaria negative test-results as confirmed by microscopy-the gold standard-should prompt clinicians to examine patients for other causes of illness and treat them accordingly [2]. However, the interpretation of negative microscopy results should take into account the high rates of anti-malarial pretreatment, which was as high as one in three admitted patients with suspected malaria in this patient cohort. A rapid diagnostic test (RDT), in addition to microscopy, could be used to detect the HRP2 malaria antigen in patients who recently received anti-malarials and whose blood films are, thus, likely to show no malaria parasitaemia [2]. RDTs can give positive results for up to 1-month after parasite clearance [2]. One in six cases of confirmed malaria, none of whom had severe symptoms, did not receive anti-malarials during the current hospitalization, which raises concern over the safety of in-patient care at this tertiary care hospital. In high transmission zones, many patients with other causes of admission could carry malaria parasites without symptoms; however, they should receive anti-malarial treatment when the infection is confirmed. Poor coordination between the laboratory and clinicians is likely to lead to missed anti-malarial treatment, which is exacerbated by high in-patient loads of up to 80 admissions in wards with official bed capacity of 54 [3]. Introducing an integrated electronic health record (EHR) system to track in-patient care could significantly improve the flow of information between different hospital departments and, in so doing, promote efficient clinical management of inpatients [12].
Seven in eight in-patients initiated on injectable AS or Q did not receive the recommended follow-up oral AL. Also, one in four in-patients who received at least one inhospital dose of prescribed anti-malarials missed the first day of their anti-malarial treatment. The missed treatment could be worsened by the observed disparities in prescribed, dispensed and administered anti-malarialssimilar to observations made elsewhere [8,10]. Possible reasons for these system lapses include; (i) drug stockouts, (ii) poor communication between clinician and patient/caregiver and, (iii) work overload [3]. The hospital should improve its stock forecasting for in-demand anti-malarials, promote intern-pharmacist-led bedside dispensing to reduce the clinicians' workload during drug administration, and improve supervision of junior and mid-level clinicians to promote accountability to inpatients and the hospital [3].
Each additional admission diagnosis increases by more than two-fold the odds of missed Day 1 dosing of prescribed anti-malarials, which underlines the need for prompt availability of malaria test-results to promote the timely initiation of anti-malarials. Prompt and complete anti-malarial treatment rapidly eliminates malaria parasites from a patient's bloodstream [13]. Patients with severe malaria should access timely appropriate anti-malarials and complete full courses of prescribed anti-malarials to promote therapeutic success, reduce malaria-related morbidity and mortality, and prevent the emergence and spread of drug resistance [7][8][9].
In-patients with an admission malaria-in-pregnancy diagnosis seemed more likely to have a microscopicallyconfirmed malaria diagnosis than in-patients with other admission malaria diagnoses. This comparative advantage at diagnosis did not translate into better anti-malarial treatment because no pregnancy-related difference was observed in the prescription, dispensing and administration of anti-malarials. Improvement in the anti-malarial medication-use-cycle should target systemic weaknesses.
Unlike Q, the hospital frequently encounters drug stock-outs of in-demand free-of-charge AS and AL, which in-patients must purchase from private community pharmacies to prevent lapses in prescribed treatment. Drugs bought from private community pharmacies are not recorded as dispensed in the hospital pharmacy registers [3], which explains why the reported number of in-patients with administered AS and AL exceeds the number of in-patients to whom these two drugs were dispensed. AS and AL are more in-demand than Q because; (i) AS is the drug of choice for its faster parasite clearance, has a less tedious administration regime, and safer profile and, (ii) AL is administered after both injectable AS and Q as the continuation of anti-malarial treatment in severe malaria [2].
Death could be attributed to severe malaria and/or to quinine-related treatment in the 88-year-old female with a single admission diagnosis of severe malaria. The caveat to this malaria-related attribution is diagnosis based on clinical judgement only (in the absence of microscopy results), unknown HIV-serostatus, advanced age, unknown random blood sugar levels and other comorbidities-especially cardiovascular comorbidities. That notwithstanding, Q was poorly administered at intervals of 23 h (between first and second doses) and 11 h (between second and third doses). Yet, 8-hourly intervals of injectable Q administration for at least 24 h are recommended until the patient is able to take oral medication [2]. The unconsciousness manifested in this in-patient is a known key sign of hypoglycaemia in severe falciparum malaria and carries a high risk of mortality [2]. Unfortunately, hypoglycaemia can result from both severe malaria and quinine-induced hyperinsulinaemia.
Thus, blood sugar levels should be checked frequently in severe malaria in-patients who receive Q [2]. Also, this in-patient had tachycardia which could have resulted from Q use and/or hypoglycaemia. With hindsight, this elderly in-patient should have been treated with injectable AS instead, although the frequent unavailability of in-demand AS, and its associated higher cost, often dictates treatment with Q. This fatal case of suspected severe malaria underpins the need for the rapid turnaround of microscopy test-results. To further improve the clinical management of in-patients with severe malaria, the hospital should also invest in routine tests for random blood sugar, haemoglobin level/haematocrit, blood gases, urea and electrolytes; and in fluid balance charts as well as intensive nursing care.
The study limitations have been reported elsewhere [3]. Briefly, the study was conducted at Uganda's National Referral and Teaching Hospital and the results might not be generalizable to facilities with lower calibres of in-patient care. Also, anti-malarials that were purchased from private community pharmacies were not documented as dispensed in the hospital register so we obtained this dispensing information by interviewing the in-patients and/or their caregivers [3].

Conclusion
Half the malaria microscopy results were not available to guide the clinical management of malaria despite that the rate of testing was high. Seven in eight in-patients initiated on injectable AS or Q did not receive the recommended follow-up treatment of oral AL. One in four in-patients delayed to initiate hospital anti-malarials by at least one calendar day. The hospital should review its workflows to encourage prompt availability of malaria test-results to promote timely anti-malarial treatment based on confirmed diagnosis as opposed to clinical judgement only. Improved stock forecasting for in-demand antimalarials, intern-pharmacist-led bed side dispensing and frequent audits of junior and mid-level clinicians could improve the quality of treatment for malaria inpatients in Uganda.