Challenges to malaria surveillance following elimination of indigenous transmission: findings from a hospital-based study in rural Sri Lanka

Abstract

Introduction

Sri Lanka has eliminated local transmission of malaria. Assessing physician preparedness for early case detection is important, in order to prevent re-establishment of local transmission.

Methods

Adherence to malaria screening practices in patients admitted with fever to 12 hospitals in a previously malaria endemic district was evaluated using a cross sectional survey. In addition, knowledge and attitudes among doctors on current malaria surveillance practices and treatment recommendations was assessed.

Results

Of 403 fever patients, 150 warranted screening for malaria under the criteria defined by the Anti Malaria Campaign (AMC), with 93 of them having fever for over 7 days. Of these eligible patients, 12.6% (19/150) were investigated by doctors (including 3 persons with fever over 7 days), 14.6% (22/150) by laboratory staff and 72.6% (109/150) by the research team. The majority of doctors were not familiar with the treatment guidelines for malaria (76.5%, 75/98).

Conclusions

Mandatory continuous medical education programmes need to continue to ensure that malaria remains on the differential diagnosis of a fever patient, especially in patients with fever over 7 days. It is essential to publicize the availability of free-of-charge malaria diagnostic facilities, and to ensure that proper notification procedures are followed when a malaria patient is diagnosed.

Introduction

The history of Sri Lanka has been intertwined with massive epidemics of malaria that killed thousands of its population.¹ The largest of them occurred in the North Western and North Central parts of the country in 1934–1935, killing an estimated 80 000 people within 7 months.^2,3 After gaining independence from the British Empire in 1948, the newly established government of Ceylon (Sri Lanka) strengthened the resettlement efforts combined with a multidisciplinary well-funded approach to tackle malaria. With the advent of new strategies, the reported number of cases of malaria in 1963 dropped to an all-time low of 17.⁴ Elimination of malaria was thought to be imminent. Increased optimism and complacency on the part of all stakeholders led to parasitological surveillance being relaxed and vector control being stopped. This along with the financial constraints that were imposed on the budget of the Anti Malaria Campaign (AMC) activities due to a decline in the number of malaria cases, allowed a malaria epidemic to sweep through the traditional malaria endemic areas of the country in 1967/69.¹

Sri Lanka is once again at a critical juncture with regard to fighting malaria. It is one of the few countries that has successfully eliminated malaria by interrupting local transmission, and will be eligible to apply for the certificate of elimination by WHO in 2016.⁵ However, there are concerns that the mistakes of 1963 may be repeated once again, with serious consequences, as many neighbouring countries still have rampant malaria transmission.

The most likely threat of re-establishment of infection comes from travel across borders. With increase in international travel, especially to and from the South East Asian countries where malaria is rampant, continued malaria surveillance is critical to diagnose cases of imported malaria early.⁶ The primary responsibility in this regard lies with general practitioners, first contact doctors and hospital physicians who encounter fever patients in large numbers. The whole surveillance system depends on them considering malaria as a differential diagnosis under appropriate circumstances, requesting screening and notifying public health officials. Even experienced practitioners can be complacent and reluctant to think of malaria as a differential diagnosis in a zero-prevalence situation. Such complacency can easily give rise to a situation where cases of malaria are missed, leading to re-establishment of locally transmitted disease, since the Anopheles mosquito that transmits the parasite is still abundant.^4,6 Therefore it is important to periodically evaluate malaria surveillance practices among physicians and assess their knowledge, attitudes and practices (KAP) regarding malaria, in order to identify areas that need continuous medical education. Such research and audits should be a standard practice for all countries entering elimination stage of this resilient infection. The aim of this study was to describe malaria surveillance practices among hospital physicians in selected hospitals of the North Central province of Sri Lanka, which was a previously malaria-endemic area when transmission was active.

Methods

Study design, study population and sample size

This was a cross sectional descriptive study carried out in 12 hospitals in the Anuradhapura district of the North Central province of Sri Lanka. This district was selected because it has reported a high incidence of malaria in the past four decades with historical records showing periodic epidemics of malaria occurring over the 20th century,⁷ it hosts the biggest field military hospital in the country and its geographical and climatic patterns also favour re-establishment of locally transmitted disease should the infection be imported.⁸ Therefore this part of the island is a key region where physicians need to keep active surveillance for re-emergence. The district serves as the bridge between the north and the south of the country. The Northern Province has close ethnic and cultural contacts with South India where malaria transmission is active.⁹

The district covers an area of 7100 km², and has a population of approximately 860 000.¹⁰ It has a number of hospitals belonging to different categories of the hierarchical system of hospitals, starting from a teaching hospital at the top to rural hospitals at the bottom. The hospitals included in this study were: the only tertiary level teaching hospital, the Victory military hospital, two base hospitals, five district hospitals and three rural hospitals with inward facilities, selected randomly and scattered throughout the district. The military hospital is the largest one serving the North, North East and Central Province, where most military personnel, who are considered as a high-risk category for malaria transmission, seek treatment. Large numbers of security personnel are stationed in these three Provinces for security purposes and to carry out rehabilitation and reconstruction work following the civil war in Sri Lanka that ended 5 years ago.¹¹

The study comprised two sections and hence two target groups of respondents. The first section studied the adequacy of vigilance for malaria by clinicians while the second component of the study was on the knowledge, attitude and practices of doctors regarding malaria.

In the first part of the study all patients with fever who were admitted to medical and paediatric wards of the selected hospitals were enrolled. The fever patients who reported were divided into two categories: fever for more than 48 hours (and less than 7 days) without any localizing symptoms or signs either clinically or by investigations; and fever without a diagnosis for 7 or more days after reasonable basic investigations. Screening for malaria was deemed necessary if a patient with fever fulfilled any of the following criteria: history of travel to a malaria endemic country within one year; history of malaria within last 3 years; anaemia and/or hepato-splenomegaly; recipients of recent blood transfusions within 3 months; fever of unknown origin or any other features of severe malaria. These criteria were based on guidelines for screening by the AMC of Sri Lanka, the official government body that oversees and coordinates all anti-malaria activities within the country, and has been functioning in this capacity for over a hundred years since its inception.¹ Sample size was calculated based on a population size of 860 000, response distribution of 50% (assuming that the prevalence of a malaria diagnostic test done among febrile patients without an obvious cause for the fever following a minimum stay of 48 hours after admission was 50%, in order to generate the largest possible sample size), confidence level of 95%,margin of error of 5%. Using standard calculators, the required sample size was 384, and anticipating a non-response rate of 5%, our sample size was determined at 403.

Data collection

Structured interviewer administered questionnaires were developed and pre-tested for both components of the study. The questionnaire to determine vigilance of medical staff regarding malaria collected patient data on demographics, risk factors, details of current febrile episode, likelihood of malaria as a differential diagnosis, treatment received, final diagnosis and activities performed with regard to malaria diagnosis by medical staff. If malaria was suspected, the following actions were expected: testing by rapid diagnostic test (RDT) and/or blood smear for malaria parasite, and notification to the AMC. If a malaria test had not been requested among patients who met the testing criteria the researchers suggested the treating physicians test the patient.

In the second component of the study, all consenting doctors serving in out-patient departments, medical and paediatric wards in all these hospitals were interviewed by four trained medically qualified persons. Data from respondents were coded to ensure anonymity. Knowledge on malaria was assessed on: the current situation of malaria transmission in the country, criteria on when to suspect a diagnosis of malaria, treatment, prophylaxis and notification practices of a potential case of malaria. The expected level of knowledge was to be aware of current epidemiological trends of malaria and to be aware of malaria treatment guidelines issued by the Ministry of Health, Sri Lanka.¹² Attitudes were tested, using a four-point Likert scale, on statements relevant to continued malaria surveillance in Sri Lanka.

Ethics approval was obtained from the Research Ethics and Higher Degrees Committee, Faculty of Allied and Medical Sciences, Rajarata University of Sri Lanka and the Oxford Tropical Research Ethics Committee (OXTREC), University of Oxford. Permission was obtained from the relevant Health Administrators of the Anuradhapura district.

Data analysis

Data were analysed using SPSS version 20 (IBM, Armonk, NY, USA). Frequencies, proportions, means and standard deviations were used for the descriptive analysis. Comparisons were evaluated using χ² test or Fishers exact test, with statistical significance set at 0.05.

Results

Malaria surveillance among patients with a febrile illness

The study was conducted in 2014 over a 3 month period. There were 403 patients (236 male [58.5%], 167 female [41.5%]), recruited from the 12 hospitals, of which 378 (93.8%) were followed up and the results of investigations and final diagnosis at discharge were recorded. Risk factors for acquiring malaria were: having received a blood transfusion within 3 months (21/403, 5.2%), and not using mosquito repellents and nets regularly (12/403, 2.9%). None of the patients had visited a malaria endemic country within 1 year prior to the onset of illness.

There were 150 patient encounters that warranted testing for malaria according to criteria defined in the Methods section. Of these, 93 patients had fever over 7 days duration. Out of these 150 patients, 114 (76.0%) had a malaria test done and recorded as negative. Doctors treating the patients spontaneously requested the tests in 12.6% (19/150) of patients (of whom three had fever over 7 days and four gave a history of a risk factor for acquiring malaria). Public health laboratory technicians working for the AMC carried out testing in another 14.7% of (22/150) patients as a part of routine fever surveillance and in the remaining 109 (72.6%) patients who fulfilled the criteria for selection, testing was done at the request of the researchers in consultation with the treating physicians. In 36/150 (24.0%) patients, test results were not available since they had been discharged following recovery or had left against medical advice. Proportion of screening by blood smears, RDTs, or both were 82/114 (71.9%), 8/114 (7.1%) and 24/114 (21.1%), respectively. In 51.7% of tested patients (59/114), the results were available on the same day, and the delay was comparatively more in the teaching hospital as compared to the other hospitals. None of the tested patients were positive for malaria. A total of 338 patients had a final diagnosis written on the Bed Head Ticket at the time of discharge, the top three being lower respiratory tract infections (94/338, 27.8%), undifferentiated viral fever (58/338, 17.1%) and dengue (37/338, 10.9%).

Knowledge and attitudes regarding malaria surveillance among physicians

A total of 132 doctors working in medical and paediatric wards and the out patient department of the selected hospitals were eligible to be included in the study, of which 98 doctors were interviewed. Approximately 8% (83/98) of the sample was less than 35 years of age, which meant they had graduated within 8–10 years and were less likely to have encountered malaria patients during training due to a low incidence of malaria that was reported in the country. The characteristics of the sample are shown in Table 1.

When the knowledge on the current situation of malaria in country was assessed (Table 2), 21% (21/98) knew that there were no indigenous cases of malaria being reported over the past one year, 50% (49/98) knew that imported malaria cases were increasing in numbers, 80% (78/98) knew that Sri Lanka remained at high risk for transmission due to the presence of vector and 97% (95/98) were aware that the country was on a course to eliminate malaria. Having a postgraduate medical qualification was significantly associated with better knowledge in this regard (p=0.00) while years of service, and age of respondents did not show a similar association.

The top five clinical scenarios where malaria would be considered in the differential diagnosis were indicated as admission of an unconscious patient (95%, 93/98), history of foreign travel plus fever (87%, 85/98), domestic travel to an endemic region plus fever (86%, 84/98), anaemia plus fever (85%, 83/98) and pyrexia of unknown origin (85%, 83/98). However, only 67% (66/98) of respondents admitted to asking about foreign travel in patients with fever while on their daily ward rounds. Despite all being aware that blood smear testing was available for diagnosis in the hospitals, many (69%, 68/98) named clinical diagnosis as the method of choice to diagnose or exclude malaria. Only 34% (33/98) knew about RDT for diagnosis.

The knowledge regarding treatment of malaria was inadequate with only 40% (39/98) and 16% (16/98) naming chloroquine alone and a combination of chloroquine with primaquine (for radical cure) respectively as the drug of choice to treat vivax malaria. Interestingly, 46% (45/98) were unaware, had forgotten or gave an incorrect answer regarding the treatment for vivax malaria. Having postgraduate qualifications or being a consultant were significantly associated with better awareness (p<0.05) in this regard. Only 14% (14/98) of the respondents correctly identified artemisinin combination therapy (ACT) as the treatment for uncomplicated falciparum malaria and only 3% (3/98) correctly identified that it should be combined with primaquine. For treatment of severe falciparum malaria, quinine was correctly named as the treatment of choice by only 21% (21/98) of respondents. Just one respondent knew that, when the patient is fit for oral therapy, treatment should be continued with ACT. Twenty three (23/98, 23%) doctors had read the national circular issued by the Ministry of Health regarding treatment guidelines for malaria.¹²

Regarding notification, 70% (69/98) knew that reporting of all diagnosed cases of malaria was mandatory, but only 32% (31/98) knew that notification should happen in all suspected cases as well. A significant 48% (47/98) of respondents did not know the correct procedure for notification of suspected and confirmed cases of malaria.

Currently two drugs, i.e., chloroquine and mefloquine, are recommended for prophylaxis of malaria by the AMC for Sri Lankans, depending on the country they are travelling to. In the sample, 61% (60/98) knew about the former, and one person knew about the latter as a recommended prophylactic agent. Only 24% (24/98) of the sample correctly identified AMC Headquarters or the Office of the Regional Malaria Officer as the place to obtain prophylactic drugs free of charge. Ninety eight percent of the sample (96/98) knew that prophylaxis is recommended to those travelling abroad to endemic countries but 76% (74/98) incorrectly believed that it is also recommended for high risk groups (e.g., pregnant mothers, security forces personnel) living in previously endemic areas within the country.

With regards to attitudes pertaining to continued malaria surveillance in the country, respondents were asked if they thought random testing at the outpatient department was a necessity, to which 15% (15/98) strongly agreed and 43% (42/98) agreed. The majority (97%, 95/98) believed that all travellers to endemic countries should be given prophylaxis and 97% (95/98) were also in agreement for reporting suspected and confirmed cases of malaria.

Discussion

Following resurgence of malaria in the late 1960s, it has been the single most important disease entity responsible for hospitalization in Sri Lanka until the mid-1990s.¹³ Thereafter, malaria continued to be a leading cause of hospital admissions until it started declining after 1999. With no indigenous malaria cases being reported since November 2012 and hospitalization for malaria being sporadic at present, the new generation of healthcare workers, especially doctors, passing from state universities have only rarely encountered a case of malaria. Interestingly, it has now become a disease learnt from textbook descriptions. This study provided an opportunity to assess the knowledge and attitudes of the doctors and their behaviour in considering malaria in the differential diagnosis of a fever patient and the current prevalence of malaria in a previously high malaria endemic area. With the threat of re-emergence a possibility due to the increasing number of imported infections,¹⁴ it is necessary to understand the physicians' attitudes, and factors that may lead to inappropriately relaxed malaria surveillance practices. Lessons learnt in this regard will be useful to many countries that are expected to enter the elimination phase in the years to come.

Sri Lanka has a well-organized parasitological and entomological surveillance system functioning at grass roots level in every administrative division of the country that was previously endemic, coordinated by the AMC.⁴ However, a similar level of coverage and quality does not appear to have reached the practicing doctors as they are less aware of national guidelines in relation to malaria, including treatment guidelines. In a profession where pattern recognition is a key to making a diagnosis, having never seen a case of malaria adds to the problem of making a timely diagnosis when confronted with a case. One way out of this is to have a systematic flowthrough chart where a patient gets categorized as a potential candidate for screening by default if certain criteria are met. This mechanism has been put in place with the AMC having specified criteria as to when to suspect a case of malaria.¹⁵

Doctors serving in hospitals located in the past hotspots of malaria transmission should be more vigilant and knowledgeable about malaria. One may argue that routine transfer of medical officers (every fourth year as per government regulations) would mean that doctors in these hospitals may not have spent a greater part of their career in previously endemic regions to be more alert to the disease. Other contributory factors maybe that they might not have participated in the in-service training programmes carried out by the AMC, the training programmes might not have been geographically extensive to cover the hospitals they worked in, or they might not have been frequent enough to compensate for the turnover of doctors. Another possibility is that the training programmes, though conducted regularly, were not effective enough to impart the necessary knowledge. Having postgraduate qualifications were the only predictive factor for having a better awareness in many of the aspects that were tested regarding malaria. Postgraduate students, by nature of their curriculum, have to read up on current guidelines on malaria which probably ensured that they were well informed compared to others.

The drugs used for the treatment of malaria are controlled substances and are only available in government hospitals and from the AMC. Treatment of malaria must be done according to the national treatment guidelines issued by the Ministry of Health in 2008.¹² Following the completion of this study, a new treatment guideline was issued by the AMC, of which the online version is freely accessible on the AMC website.¹⁵ The study demonstrates that more effective measures of communication on malaria to health care providers are required. Recently, it has been demonstrated that small group, interactive, seminar type health education programmes were very effective in knowledge transfer regarding malaria with respect to a high risk group for transmission: the security forces. Despite many of them not having any medical background, most showed a significant improvement in knowledge and attitudes for better malaria surveillance in the post intervention assessments.¹¹ Though such an intervention could be time and resource consuming, it can also be an investment for the future to avoid re-emergence of infection. Yet a major challenge in conducting similar training programmes for the clinicians would be the difficulty in gathering them to a single location during working hours, due to their high workload. Furthermore, targeting them for a training programme after working hours will be less feasible as a majority engage in private medical practice.

In this KAP study, the knowledge and attitudes components of doctors were assessed via questionnaires. However, the practices component was actually observed in the wards as to how they respond to situations that warrant malaria screening. In order to allow adequate time for the clinicians to reach a provisional diagnosis, patients who had been admitted for less than 48 hours were excluded from the study. Still, malaria was not in the differential diagnosis, even of a patient presenting with over 7 days fever, as only three out of 93 patients had been investigated for the disease. As malaria is fast becoming a ‘forgotten disease’ among the doctors, in a majority of cases the research team mediated and alerted the practicing doctors to request a test when they did not spontaneously do so.

It is interesting to note that when asked whether fever patients attending the outpatient department of a hospital should be randomly screened for malaria, a significant proportion of respondents disagreed. Testing for malaria is not an expensive, cumbersome test. All that is required is a trained microscopist and a microscope (which all the hospitals in the study were equipped with). Test results can usually be obtained in a few hours unless there is no dedicated microscopist in the hospital in which case the test is performed in the hospital's pathology department, which also handles a large number of other investigations. This form of surveillance (activated passive case detection), where patients with fever who present to medical institutes are screened for malaria, is an accepted mode of screening with better ability to detect new cases, especially when a country has reached elimination stage as Sri Lanka has done.^16,17 It is interesting to note that a significant proportion of respondents perceived this form of surveillance negatively. Perception that malaria is not a healthcare problem that needs such extensive scrutiny, increased patient waiting times, incorrect perception of costs involved, increased workload in already overcrowded outpatient departments, or just simple unawareness of benefits of activated passive case detection could be the possible reasons. In addition, in a state of extremely low or no prevalence, there may be significant inertia on part of treating physicians to request a test for malaria. When pressed to make a hurried diagnosis they are more likely to think of more prevalent life threatening infections such as dengue or leptospirosis, than malaria. The problem also has to be considered in the current socio-cultural context where the local media and populace is seriously concerned about infections such as dengue and physicians may be subconsciously biased to exclude the ‘politically correct’ infections first to avoid allegations of misdirected care. They may also be biased into thinking that junior doctors and colleagues may question the consideration of a rare infection rather than a common prevalent infection. Exploring all these hypotheses are beyond the scope of this study but need to be addressed in a qualitative study involving focus group discussions. Findings of such an exercise will yield important information on tackling physician inertia to improve malaria surveillance.

Limitations

There are 19 districts in the country extending across eight provinces that were previously endemic for malaria. This study evaluated only one district which would limit the generalizability of the present findings. Only the hospitals with inward facilities were included in the study as not only would patients in rural areas with unresolving fever seek treatment in state hospitals with inward facilities but clinicians practicing in these areas having detected ongoing fever without an obvious cause would have admitted them for further investigations.

In the current context, where malaria has become a rare disease, the probability of a malaria case being included in a sample of 150 patients eligible for screening is extremely low. There would have been a higher probability of diagnosing a patient with malaria should a larger sample size have been included and if more stringent criteria had been followed in recruitment of individuals for the study, e.g., limiting the study to patients who fell within the criteria defined by ‘pyrexia of unknown origin’ (fever more than 3 weeks) instead of taking patients with fever of over 48 hours without an obvious cause. This would have resulted in a better yield as the majority of fever cases in the latter group could have been due to other common causes of fever such as viral fevers. Given the likely consequences of a missed malaria case and also the relatively low costs of screening, inclusion of patients with over 48 hours of fever allowed the investigators to get a better understanding of the tendency of clinicians to suspect malaria in fever patients, and whether they inquire about the risk factors for malaria and request for screening for such patients.

The study did not include the obstetric wards and their doctors in any of the hospitals studied, hence the KAP of doctors attending to pregnant women, which is a high-risk group, were not evaluated.

Conclusions and recommendations

In this survey of 12 hospitals in Anuradhapura district which was previously endemic for malaria transmission, malaria surveillance alertness of treating doctors with regard to patients with undiagnosed fever could be improved. Despite the availability of rapid screening services both in the hospitals as well as with the Regional Malaria Officer, a request for screening was not made spontaneously on most occasions, even in patients with fever over 7 days. The knowledge on treatment guidelines for malaria could be improved amongst the clinicians by carrying out more focussed awareness programmes by the AMC in an attempt to promote early diagnosis of malaria and prevent the disease being re-introduced to the country.

For reasons that were not explored here, the support for activated passive case detection among patients attending out-patient departments were not convincing. Apart from the treatment guidelines, knowledge on other aspects of malaria transmission and surveillance was significantly better among those that had postgraduate qualifications, showing the benefit of continuous professional education. The changing dynamics of physician attitudes towards malaria in Sri Lanka after reaching elimination stage needs urgent attention and this study can be a model to explore similar situations in other countries entering the elimination phase. Not addressing the issues of complacency in health care providers after elimination of local transmission can easily reverse the ground gained against malaria if imported cases re-establish transmission aided by failed surveillance.

We therefore recommend: to re-evaluate the AMCs training plan for clinicians in terms of its coverage and quality. Considering a training of trainers model may be worthwhile to reach a wider target population distributed in a wide geographical area, within a limited period of time, given the importance of achieving the minimal knowledge required for the sustenance of malaria elimination status; to re-evaluate the undergraduate teaching and the in-service training activities with regard to malaria with a view of improving their long term effectiveness; to analyse the participants profile attending awareness programmes that are currently being carried out for health care personnel by the AMC in an attempt to establish whether the relevant target groups are adequately covered, e.g., pre-registration house-officers, non-specialist-training grade doctors, specialist trainee doctors, and consultants.

Authors' contributions: SDF, SR and RGP conceptualized the study. SA and SRJ completed the data collection.SA did the first analysis. CR repeated the analysis and wrote the first draft of the manuscript. All authors contributed to the final manuscript and read and approved the final version. SDF is the guarantor of the paper.

Acknowledgements: The authors would like to thank the staff of the Regional Malaria Office, Anuradhapura for the support received towards carrying out this study and Prof. Sisira Siribaddhana, Dean of the Faculty of Medicine, University of Rajarata for his continuous support. The authors also thank the Medical Directors of the selected hospitals and Consultants/Medical Officers in charge of the wards for all the assistance provided.

Funding: Financial assistance was received from the Nuffield Department of Population Health University of Oxford, Global Health Policy Programme at Green Templeton College, Oxford and National Science Foundation, Sri Lanka [RG/2014/HS/03].

Competing interests: None declared.

Ethical approval: Ethical approval was obtained from the Research Ethics and Higher Degrees Committee, Faculty of Allied and Medical Sciences, Rajarata University of Sri Lanka and the Oxford Tropical Research Ethics Committee (OXTREC), University of Oxford.

References