Background
The last 5 to 10 years has seen a substantial increase in the availability and use of malaria Rapid Diagnostic Tests (mRDTs) in many parts of the world [
1]. Consistent with these global trends, Papua New Guinea (PNG) with support from a Round 8 Global Fund grant has procured over three million mRDTs since 2010 [
2]. This supports a revised national malaria treatment policy (NMTP) stipulating that all fever or suspected malaria cases be tested for malaria infection by microscopy or mRDT [
3]. Internationally, the scale-up of mRDT availability has often led to substantial reductions in antimalarial prescription as health workers shift from a presumptive to a ‘test and treat’ malaria case management approach [
4‐
6]. There is some evidence to suggest a similar change in health worker practice is occurring in PNG. A recent study found that antimalarial prescriptions to febrile patients declined from 96.4% of cases before the introduction of the test and treat policy to 39.0% in the 12-month period immediately post-implementation [
7]. The availability of mRDTs increased from 8.9% of surveyed facilities to 53.4% across the same time period.
Whilst mRDTs can confirm or rule out malaria infection, they provide no support to health workers for diagnosing and treating non-malarial fevers. Fever is an unspecific syndrome and diagnostic tools to assist in the accurate identification of its aetiology in mRDT negative patients are scarce in resource-poor settings [
8]. The historic practice of presumptive malaria diagnosis may further undermine the quality of non-malarial febrile case management as the health workforce has little acquired experience in the clinical diagnosis and treatment of non-malarial febrile patients [
9]. Many febrile patients attending PNG health facilities test negative for malaria infection and the malaria burden is declining in the general population [
10,
11]. Thus, the health workforce is increasingly, and quite suddenly, required to respond to febrile illnesses shown to be non-malarial in origin via mRDT.
A health worker training program was carried out in PNG pre-implementation of the new NMTP. This program provided detailed instruction on mRDT use and strongly emphasised the importance of restricting antimalarial prescription to test-confirmed cases [
12]. Nevertheless, the training program provided relatively little detail on non-malarial febrile case management beyond reference to an Integrated Management of Childhood Illness (IMCI) flowchart. No instruction on the management of non-malarial febrile illness in adolescent or adult patients was provided. The PNG National Department of Health and the Paediatric Society of PNG issues standard treatment guidelines which outline recommended management of a range of common illnesses, including non-malarial febrile illness [
13,
14]. Apart from these reference manuals, no other training, resources or support have been specifically provided to assist the PNG healthcare workforce manage the increasing number of non-malarial febrile illness cases identified in the context of the new NMTP. Accurate diagnosis of non-malarial febrile illness is further comprised by the absence of on-site laboratory support at most primary health care facilities in PNG [
15]. The general lack of curative treatments for most viral febrile illnesses may be an additional obstacle to implementing accurate evidence-based treatment.
This paper examines how the PNG primary healthcare workforce diagnosed and treated non-malarial febrile patients in the period immediately following the widespread introduction of mRDTs. The primary research questions included: what percentage of non-malarial febrile patients are provided a recorded diagnosis? What diagnoses are being recorded? What medications are being prescribed? And to what extent do the medications prescribed adhere with national standard treatment guidelines?
Discussion
This paper presents data pertaining to the clinical case management of non-malarial febrile patients in the 12-month period immediately following the discontinuation of a treatment protocol in which most febrile patients were presumptively treated with antimalarials. Data were derived from longitudinal surveillance in seven outpatient sentinel health facilities and from a cross sectional survey in 36 health centres across 17 provinces in PNG.
Findings from the OS sample indicate that the vast majority of non-malarial febrile patients had a formal diagnosis recorded, in the CS sample this was little more than 50%. As the latter comprised a greater number of health workers, and as these health workers were not prompted to provide a diagnosis (as was the case in the OS sample), then the CS percentage may be the more accurate reflection of standard practice. Respiratory tract infections, often pneumonia, were the most common diagnosis provided across both datasets. Diagnoses of diarrhoea and general body aches were also prominent across samples, although at a substantially lower frequency. Respiratory tract infections and diarrhoea have historically been reported at a high frequency on the PNG National Health Information System [
17] and are ranked highly in PNG burden of disease estimates [
18]. Thus, the specified diagnoses of non-malarial febrile illness are largely consistent with existing data sources.
It was not possible to determine whether the recorded diagnoses were accurate or not in the context of this study, with the exception of malaria and anaemia. However, the small number of studies that have examined the aetiology of non-malarial febrile illness in PNG have identified dengue as a cause in approximately 10% of cases [
19,
20] and a chikungunya virus outbreak was detected in PNG at the time the reported data were collected [
21,
22]. A recent review of non-malarial febrile illness in the neighbouring South East Asia region further identified dengue, typhus and
Leptospira spp. as the most commonly reported pathogens across 146 studies meeting the inclusion criteria [
23]. As neither dengue, chikungunya, typhus or leptospira spp. were diagnosed in a single case in either sample, as malaria was diagnosed in between 8.9 to 18.8% of mRDT-negative cases and as anaemia was substantially underdiagnosed despite the availability of a Hb-measurement in the OS sample, there is reason to question the accuracy of recorded diagnoses. The limited diagnostic/laboratory support available in primary health care facilities in PNG restricts the degree to which health workers can investigate the causes of febrile illnesses and may consequently influence the accuracy of diagnoses provided. The threat of misdiagnosis may, therefore, be considered a result of limitations in the broader healthcare system rather than health worker ability per se.
Prescription patterns were consistent across datasets. Over 90% of patients in both the OS and CS sample were prescribed one or more medications, most commonly some form of antibiotic, analgesic and/or anthelminthic. Antimalarials were provided in 18.8 and 13.2% of OS and CS cases, respectively, despite the negative mRDT result. This rate of antimalarial prescription to mRDT negative patients is lower than that reported in other settings [
24‐
26] and represents a substantial reduction in the rate of antimalarial prescription to febrile patients compared to the practice observed prior to the change in treatment protocol [
27]. Thus, while not fully compliant with the revised protocol, PNG health workers have seemingly made radical and appropriate adjustments to their antimalarial prescription practices in a relatively brief time frame and perhaps at a faster rate than their international peers in similar circumstances.
The adherence analysis indicated that the prescription provided was consistent with that recommended in the respective PNG standard treatment manual for the specified diagnosis in fewer than 20% of cases (OS sample only). The majority of prescriptions (56.3%) were rated ‘partially adherent’, indicating that at least one, but not all, of the recommended medications was provided or the correct class of medication was provided, but not the recommended drug. It was not possible to determine what influence the availability of the respective medications played in this outcome. This result, therefore, most likely reflects limitations in both medication availability and health worker practice, yet it remains the case that the medications provided were rarely in full accordance with those recommended for the diagnoses given. A previous study, not restricted to mRDT negative cases, reported ‘appropriate’ prescription rates (based on diagnosis) of between 62.1–69.6% in two secondary care settings and an outpatient facility in PNG [
28]. Thus, the relatively poor adherence to prescription guidelines reported in this study suggests that PNG health workers may be less likely to provide a recommended prescription when treating non-malarial febrile illness patients. However, in the absence of an expert reference diagnosis in the context of this study, the extent and consequences of any potential misdiagnosis and non-recommended prescription cannot be judged.
Previous studies have reported similar rates of antibiotic prescription to mRDT negative patients as reported here, from 61.4% in Uganda [
29] to 78% in Tanzania [
30]. Accordingly, the rates may be considered somewhat normative in a low income country setting in which a shift from presumptive to parasitological confirmed malaria diagnosis has recently taken place. Nevertheless, the rate of antibiotic prescription is high and raises the possibility that primary health care workers in PNG may be substituting antimalarial medication with antibiotic medication in response to the revised NMTP, as has been previously suggested [
6,
9]. The potential over-prescription of antibiotics cannot be reliably established in the absence of an expert reference diagnosis including, in some instances, additional diagnostic testing; nevertheless, a recent study reported unnecessary antibiotic prescription in 29% of 6969 observed illness episodes in outpatient health services in PNG [
31] which is somewhat consistent with the rate of antibiotic over prescription (41%) reported herein. Furthermore, an analysis of medication prescription to malaria mRDT
positive patients in the CS survey [
7] found that out of 54 mRDT positive patients, 98.2% were prescribed an antimalarial, yet only 14.8% were prescribed an antibiotic. This would indicate that health workers are selectively (and often unnecessarily) prescribing antibiotics to mRDT negative cases as opposed to routinely providing them to all febrile patients.
This study was not without limitation. Firstly, the fact that (to a large extent) the accuracy of the reported diagnoses was not able to be assessed in either the OS or CS sample remains an important knowledge gap. Whilst assessing the accuracy of non-malarial febrile illness was not an aim of the study, without this essential information it is difficult to reliably assess the quality of health worker practice. A further limitation is that the study did not take into consideration medication supplies at the respective health facilities in either survey. The supply of anti-malarial drugs or other medications could have an implied effect on the prescribing behaviour of health workers [
32]. In addition, health workers in the OS sample were required to complete a diagnosis and medication section on a research-specific case report form which is not reflective of normal practice and health workers in the CS sample were actively observed during their respective febrile case consultations by a member of a research team. This, too, may have influenced health worker practice. The health workers in both surveys were predominantly nurses and community health workers. Therefore, the conclusions drawn concerning health worker behaviour are not a complete depiction of all health worker cadres in PNG.
The respective datasets also had their inherent weaknesses. The OS analyses were based on a large number of patient cases, yet they had been collectively treated by a relatively small number of health workers. Conversely, the CS analyses were based on a smaller number of patients, yet they had been treated by a larger number of health workers from a more diverse array of health facilities. Triangulating findings from the two datasets, as was done here, overcomes their respective limitations in part. Despite these limitations, the collective sample size of both patients and health workers, the relatively robust sampling and study protocols, the geographical spread of participating health centres and the general agreement between the two datasets with respect to reported diagnosis and prescription practice suggests the reported findings could reasonably be generalised to health centres across PNG.
Conclusions
The findings indicate health workers predominantly adhere to mRDT negative results, determine alternative diagnoses in most cases and provide medications other than antimalarials. However, the findings further suggest that a large proportion of non-malarial febrile patients are not being provided a diagnosis, the diagnoses that are given may not necessarily be accurate, the medications prescribed rarely fully adhere to those recommended for the specified diagnosis and that antibiotics are likely to be overprescribed. All of these factors strongly indicate a need for intensive and continuing health worker support in the diagnosis and management of non-malarial fevers and for thorough scientific investigation and reporting on the aetiology of non-malarial fevers across the various regions of PNG.
Interventions based on training and/or the provision of rapid diagnostic tests have demonstrably improved non-malarial febrile case management [
33], and interventions proven successful in improving malaria case management, such as text message reminders [
34], would seemingly be easily transferable. General health systems strengthening in low income country contexts such as PNG is equally essential to sustained improvement in health worker performance [
35]. As multiple support and supervisory mechanism are typically required to improve health worker practice [
36], strengthening non-malarial febrile case management in primary health care contexts in PNG will likely require a package of health worker and health system strengthening interventions, variously focusing on: further reducing antimalarial prescription to malaria mRDT negative patients; encouraging formal diagnosis and the use of clinical and available laboratory resources to inform diagnosis; rational use of antibiotics; and adherence to recommended treatment guidelines when a diagnosis is made.
Acknowledgements
The authors gratefully acknowledge the many health workers and patients involved in this study. The generous support of the respective provincial and district health authorities and the PNG National Department of Health are also acknowledged as is the PNG Malaria Technical Working Group and the many staff members of the Papua New Guinea Institute of Medical Research who assisted with data collection and management.