Findings from this study indicate that microscopy or RDT confirmed malaria diagnosis was the exception in PNG under the former treatment protocol, collectively occurring in less than 20% of observed malaria case management consultations. Microscopy or RDT was only available in 22% of the surveyed health facilities, which clearly limited their use, although even in those facilities with these diagnostic tools available only 40% of fever patients were tested. Patients older than five years of age were more likely to be tested for malaria infection relative to younger patients, although still fewer than 50% of adult (16 years +) patients were tested in facilities with the resources to do so. The study findings also suggest that clinical diagnosis was often a far from exhaustive process. Questions that could reasonably be expected to be a mandatory component of a thorough clinical assessment were often not asked and procedures such as palpating the abdomen or examining the patients’ eyes or palms were rarely conducted. Thus, it would appear that most malaria diagnoses in PNG were made presumptively, simply on the basis of the presence of fever without a thorough clinical examination and use of diagnostic tests. Anti-malarial provision to fever patients was near universal with 96.4% of the observed fever cases receiving a prescription, including the 41/50 patients who tested negative for malaria infection by RDT or blood slide.
These findings are of concern from a treatment perspective as evidence suggests many (and in some cases most) fever patients in PNG, when tested by microscopy or RDT, do not have malaria infection . Numerous patients, therefore, are likely to be receiving anti-malarials unnecessarily and, as a result, may have experienced a delay in appropriate diagnosis and treatment response. The consequences of malaria misdiagnosis have not been well examined in PNG, although international evidence suggests it may contribute to ongoing ill health and economic hardship, especially amongst the poorest members of a community [20, 21]. Malaria misdiagnosis further contributes to the development of parasite resistance to anti-malarial drugs, a significant problem in PNG at present  and an issue of global concern with respect to containing parasite resistance to the new artemisinin-based anti-malarial drugs [23, 24]. A malaria misdiagnosis also incurs economic costs in the form of unnecessary medication prescription, an issue of concern in a developing country.
These findings highlight the substantial changes that the revised PNG national malaria treatment protocol will require in terms of malaria case management practice. In particular, the proposed shift to routine testing of all fever patients by RDT or microscopy and the prescription of anti-malarials to test positive cases only are likely to challenge currently entrenched practice. The emphasis on thorough patient counselling in the new treatment protocol is also likely to take some adjusting to given the practices observed in this study. For example, the purpose of the prescribed medication was not explained to patients in 36% of cases, dosage/regimen instructions were not provided in 25% of cases, the possibility of adverse effects and what they might look like were virtually never discussed, and instructions on when to return to the health facility (if needed) were only provided in 28% of cases. On a more positive note, 79% of prescriptions conformed to current treatment guidelines and the prescription of mono-therapies was relatively rare (68/440 prescriptions). These latter findings indicate most clinicians are generally aware of and comply with recommended prescription practices and may continue to do so with the introduction of a revised protocol. Conversely, these findings also indicate that outdated mono-therapies were provided in over 15% of cases and the treatment guidelines were not followed in 21% of cases. Thus, full adherence to the new protocol in the short- to mid-term is probably unrealistic if the same has not been achieved with the former, familiar protocol.
Realistically, the successful transition from the former to the revised national malaria treatment protocol, in terms of health worker compliance, is likely to take an extended period of time; all the more so given the degree of change in clinical practice that will be required. The experience from other developing countries that have previously revised their national treatment protocols in a manner consistent with that proposed in PNG is instructive in this regard. In Kenya, in the first year post-implementation of an ACT-based national malaria treatment protocol, ACT was prescribed to fewer than 30% of fever patients . Four years later the percentage of fever patients prescribed ACT in Kenya had risen to over 60% . Consistent with the Kenyan experience, a recent review of health worker prescribing practices identified that compliance with ACT prescribing guidelines increases the longer guidelines have been in place . However, it was notable that even in health facilities in which ACT was in stock, the prescription of ACT to fever cases was below 70% in every study included in the review and typically below 50% if the study was conducted shortly after implementation of the ACT-based guidelines. Findings pertaining to RDT use depict a similar scenario. Health worker compliance with the use of RDT kits to test for malaria infection has been low , although not always , in the period immediately following their introduction. A high level of health worker compliance with RDT results also appears difficult to achieve with many studies reporting anti-malarial prescription to between 30 to 50+ % of test negative cases [27–29], even when the testing programme has been in place for some years .
What these studies collectively suggest is that achieving a high level of health worker compliance to a revised RDT/ACT-based malaria treatment protocol takes a number of years. A high proportion, possibly a third or more, of test negative cases are still likely to be prescribed anti-malarials some time after the introduction of RDT kits and it is unlikely that more than 70–80% of test positive cases will be prescribed the correct ACT even two to three years post-implementation of a revised protocol. Progress in PNG is unlikely to be any faster given the level of change in malaria case management practice that will be required. The potential barriers to health worker compliance with malaria treatment protocols are multiple, ranging from supply problems to peer/patient pressures to insufficient and/or inappropriate training/support to mistrust in the new medicines or diagnostic resources [17, 18, 30, 31]. A small number of studies have sought to identify interventions that might usefully improve adherence to malaria treatment guidelines, although often with limited success. For example, the impact of a three-day, in-service training and the provision of various resources and job aids on malaria case management were modest at best in a pre-/post-intervention study conducted in Kenya . The authors subsequently concluded that one-off training interventions, even when supported by training materials and job aids, are unlikely to be effective if follow-up support and supervision is not provided. Reflecting the benefit of regular longer-term support, a recent study demonstrated a substantial improvement in health worker adherence to malaria treatment guidelines via the provision of regular text message reminders . Ten discrete text messages, each describing a recommended malaria case management practice, were variously sent to participating health workers personal mobile phones twice a day, five days a week over a six-month period. A 24.5% improvement in malaria case management practice (based on adherence to national treatment guidelines) was subsequently observed in the intervention group versus the control group six months post intervention. Despite this promising result, there remain few interventions that have been reliably demonstrated to improve malaria case management practice  and although regular supervision and follow-up support (via any medium) may improve performance, complete or even substantial adherence to a revised malaria treatment protocol will still likely require an extended time period.
The study presented in this paper was not without limitations. The final sample size, and especially the number of aid posts included in the sample, was lower than anticipated. This was largely due to the inaccessibility of aid posts or, more frequently, the absence of any functional aid post to survey. The study was conducted during a period of low malaria transmission (June-November, 2010) in those provinces with seasonal variation. Thus, the number of malaria patients presenting to health facilities and the subsequent pressure on staff and resources (e.g. RDT kits, anti-malarial medication) may have been lower during the survey period as opposed to peak transmission periods. Clinical observations may also have been subject to some form of bias given that participating clinicians were aware that they were being observed and their practice assessed. Any such bias was likely to have been in the direction of promoting a more thorough or accurate (according to current guidelines) clinical case consultation, although possible anxieties associated with the knowledge that one was being assessed may have negatively impacted on clinical performance in some cases. In health centre settings this source of potential bias was hopefully minimized given the duration of the observations (five days), although it cannot be discounted completely. The reported percentage of ‘correct’ anti-malarial prescriptions may be an overestimate as information on the severity of illness (uncomplicated vs. severe malaria) was not available. However, 96.1% (423/44) of patients included in this analysis were sent home at the conclusion of their clinical consultation suggesting uncomplicated malaria was the most likely diagnosis. Similarly, the assessment of whether an anti-malarial prescription was provided in the correct dosage (or not) was based on patient age as opposed to weight. Finally, the sample size employed in the regression analysis presented in Table 4 was low and the reported findings should be considered highly tentative.