Quality of malaria services offered in public health facilities in three provinces of Mozambique: a cross-sectional study

Malaria transmission occurs throughout Mozambique, ranging from low and seasonal transmission in the south to some of the world’s highest transmission rates in the holo-endemic centre and north of the country [1]. Malaria control in Mozambique is based on the pillars of prevention, including vector control and prevention of malaria in pregnancy, prompt diagnosis and provision of treatment with efficacious anti-malarials and a strong malaria case surveillance system. Malaria care is delivered primarily through public health facilities and a network of community health workers (CHWs). In 2008, the National Malaria Control Programme (NMCP) rolled out a policy of evaluating all malaria suspects (defined as individuals with malaria symptoms such as fever, headache, tiredness, sweating, chills, muscle pains and general malaise) with either microscopy or rapid diagnostic tests (RDTs); in 2017, case management guidelines were revised to explicitly recommend malaria testing of all febrile patients [2]. A cascade training on these new guidelines was rolled out through most of the country beginning with Zambézia in mid-2018, but by April 2019 had not reached all provinces.

In Mozambique, malaria testing is done using RDTs that are available at all levels, including at the community, or microscopy, which is limited to hospitals and certain health centres with laboratory capacity. First-line treatment in public health facilities is artemether-lumefantrine (AL) for uncomplicated malaria cases, oral quinine for pregnant women in their first trimester, and injectable artesunate for all severe cases. AL and pre-referral rectal artesunate are used for treatment of cases at the community level. All four age/weight formulations of AL should be available at all levels [2]. Malaria commodities are part of the essential commodities that flow through a pull-based system in public health facilities and all malaria services are provided free of charge. Nevertheless, stock-outs at the facility level remain a perennial challenge.

All-cause, under-five child mortality in Mozambique has fallen substantially in recent years, declining from 233 deaths per 1000 live births in 1990 to 97 in 2011 [3, 4]. Moreover, household survey data have shown substantial progress in the proportion of children with fever who received a finger or heel prick (from 56% in 2011 [4] to 69% in 2018 [5]) and in the proportion of children receiving an anti-malarial who received artemisinin-based combination therapy (ACT) (from 60% in 2011 [4] to 99% in 2018 [5]). Additionally, the number of individuals dying from malaria in public health facilities has consistently declined over the years from 3245 in 2014 to 968 in 2018 [6, 7].

In addition to periodic household surveys, the NMCP monitors the provision of malaria care through analysis of routine, monthly data submitted by health facilities and CHWs through the health management information system (HMIS). Analysis of these data show that 42% of persons presenting for outpatient consultation in 2017 were tested for malaria (suspect cases) and that 24% had a positive malaria diagnosis (confirmed malaria cases).

Aggregation of individual case data is useful for monitoring malaria trends and identifying data quality issues, but it is not possible to conduct a comprehensive evaluation of case management practices solely using routine data. To complement routine monitoring, health facility surveys can rigorously evaluate malaria care delivery and have commonly been used by malaria control programs in various settings [8‐24]. Researchers in Mozambique completed a retrospective analysis of routine patient register data in 2011, a few years after the introduction of RDTs and of AL; the analysis found that 72% of patients with negative RDT results received treatment [25]. Mozambique has subsequently provided additional national level training and supervision of health care providers, but little data exist to demonstrate whether case management has improved over time.

The NMCP in Mozambique implemented a health facility survey in 2018 to evaluate the quality of malaria case management and assess case management practices.

A total of 117 facilities (39 in each province) were visited during the study, resulting in 1840 outpatient and 319 HCW interviews (Table 1). The majority (72%, 1325/1840) of surveyed outpatients presented with febrile illness, with fever rates ranging from 63% in Maputo to 78% in Zambézia (Table 1). After exclusion of fevers due to malaria, between 79 and 83% of children < 5 and 46–61% of older children and adults had non-malaria febrile illness. The proportion of febrile malaria infections ranged from 4% in Maputo to 42% in Zambézia. There were no afebrile RDT+ outpatients in Maputo, but 6% of patients in Zambézia and 3% in Cabo Delgado were afebrile but tested positive by RDT during the re-exam. Fever was the main motivator of care-seeking in 38% (95% Confidence Interval [CI] 31–44%) of patients in Maputo Province, 40% (32–49%) of patients in Zambézia, and 55% (47–64%) of patients in Cabo Delgado.

Despite making up the large majority of outpatients, less than half of suspect malaria cases were found to be appropriately managed for malaria in public health facilities in three provinces of Mozambique. Even after exclusion of fevers due to malaria, the majority of patients had a febrile illness, confirming studies showing fever to be highly prevalent in outpatients throughout sub-Saharan Africa [27]. Despite this preponderance of measured fever amongst outpatients, measuring temperatures was rare and malaria testing of fever cases was sub-optimal in all provinces, below 70% in high-transmission Zambézia and Cabo Delgado, and only 33% in low-transmission Maputo.

High availability of RDT or microscopy and high rates of identification of fever cases suggests that clinicians are making the decision to not test patients that they know have a febrile illness. Non-testing was the main contributor to inadequate management of malaria cases in Maputo and Zambézia, and an important contributor in Cabo Delgado. This reflects findings from previous health facility surveys in other settings that have shown the testing step in the malaria case management pathway to be a critical barrier to correct management [8, 9, 13, 18].

Case management of malaria is a complex multi-step process, and like other similarly complex interventions, requires high performance at each step to obtain high overall coverage [28]. The quality of malaria case management was particularly low in Maputo, where only 29% of suspect cases were correctly managed. Given that Maputo has the lowest transmission levels in the country [5], this low testing could reflect HCW perceptions of the risk of malaria, low rates of supervision and training, or gaps in HCW knowledge. This matches findings from other surveys showing lower rates of testing and correct case management in low-transmission areas [8]. This finding highlights the need for continued and intensified provider training on malaria case management as geographic areas transition from higher to lower malaria burden, particularly if regional elimination goals are to be met. Although overall malaria case management performance was better in Zambézia and Cabo Delgado, where malaria cases represent a higher proportion of all-cause consultations, there were still substantial weaknesses in malaria case management in these provinces. An additional gap in the management of fever was the treatment of malaria-negative cases with anti-malarials, which ranged from 8% to 22%. While it is alarming that a large proportion of patients are unnecessarily consuming ACT, this still demonstrates important improvements from 2011 when 72% of patients with negative RDT results received treatment [25].

Notably, these findings suggest counseling of patients prescribed ACT was not effective, with approximately 40% of patients in each province leaving the health facility not knowing how many tablets to take over how many days. Moreover, the majority of patients prescribed AL, the predominant artemisinin-based combination in use in Mozambique, did not recall being told of the need to take it with food or milk. Administration of AL with a fatty meal is crucial to maximize absorption of the lumefantrine component [29, 30], and not doing so risks inadequate clearance of parasitaemia and recrudescence. Additionally, failure to tell patients to immediately return to the health facility should symptoms not improve is a missed opportunity for improving malaria case outcomes. Thus, failure to appropriately counsel patients with malaria is an important gap in quality malaria case management. Factoring poor comprehension of the dosing schedule into the calculations of the coverage with adequate treatment in true malaria cases reduces the estimate of coverage to 30% or below in each province.

Importantly, many of the main conclusions drawn from this survey can be determined through analysis of the routine data. For example, the substantial under-testing in Maputo documented during the survey is consistent with the data reported by the province through the HMIS. In 2017, Maputo health facilities reported suspect cases to be 20% of all-cause patient consults. Because data from exit interviews showed that 63% of outpatients in Maputo present with fever, a rough upper bound for the testing rate of febrile cases is 32%, which matches the testing rate of 33% observed during this survey. This finding suggests that routine data can be used to make useful inferences about testing rates with high frequency and granularity [27]. There is, nevertheless, still value in periodic health facility surveys to complement routine data analysis in order to evaluate essential aspects of appropriate case management such as quality of patient counseling, non-testing of fevers, and appropriate treatment dispensing.

Limitations related to this study design may have affected the study results and their generalizability. Exit interviews minimize the risk of bias due to the Hawthorne effect, but are subject to bias due to patient recall [31]. Some of the findings of substandard counselling quality could be due to poor recall or inadequate understanding of the survey questions. Additionally, the cross-sectional design of the survey precludes extrapolation of the findings to other time periods, and it is not known how the case management practices documented here would change later in the transmission season when malaria risk is lower or in reaction to changes in commodity availability. Lastly, this study was designed to determine the quality of malaria case management for the general outpatient population and, as such, does not provide insights into the quality of case management for severe malaria or for important sub-groups such as pregnant women as only 17 pregnant women were included in the random selection. These are important areas for future investigation to better guide malaria case management supervision, training and commodity supply systems.

Nevertheless, the results of the risk factor analysis suggest that malaria case management would improve with better availability of commodities and additional trainings and supervisions, as each of these were significantly associated with appropriate malaria case management in the risk factor analysis. Future trainings might benefit from an increased focus on universal testing of all fever cases and better counselling. However, trainings should be complemented by other activities as malaria case management and counselling was poor even in Zambézia, which had high rates of HCWs who had undergone trainings. This underscores the importance of combining training with other interventions [32] such as continued supervision post-training to reinforce implementation of best practices. This supervision would benefit from a patient observation component, where testing and counselling practices can be directly and systematically assessed.

This study of the quality of malaria case management in Mozambique found that while there have been important improvements made over time, critical gaps remain along the malaria case management cascade in Mozambique. Results varied across the different provinces included in this study, but failure to test all febrile patients and provide appropriate counseling was consistently problematic. These findings underscore the need to strengthen provider testing of all patients with fever, improve provider adherence to negative test results, and enhance effectiveness of counselling of patients across epidemiological settings in Mozambique.