The most important lesson learnt from the severe acute respiratory syndrome (SARS) epidemic was that ‘inadequate surveillance and response capacity in a single country can endanger national populations and the public health security of the entire world’ (Heymann & Rodier 2004). Regrettably, developing countries with the greatest risk for epidemics of communicable diseases also generally have environments most conducive to disease transmission, particularly by faecal-oral route. They often lack the capacity to promptly detect and adequately respond to disease outbreaks. This has historically characterized many Pacific Island Countries and Territories (PICT), where large-scale migration considerably increases the risk of disease importation. Recognition of this vulnerability and the need to confirm the adequacy of acute flaccid paralysis (AFP) for regional polio-free certification resulted in the establishment of a hospital-based surveillance system for AFP, measles and neonatal tetanus in 1997, with 53 hospitals and most PICT participating (O'Leary 2000). Sustained efforts have been made to strengthen Western Pacific regional and national outbreak capacity through the development of the Pacific Public Health Surveillance Network under the auspices of the Secretariat of the Pacific Community and WHO, with attempts made to harmonize health data needs and surveillance; to develop supportive computer applications; to train national-level staff in appropriate field epidemiology and public health surveillance skills; to promote the use of email sharing of information; to publish regular health bulletins; and to build laboratory connectivity (O'Leary 1998; Souares 2000).
The focus of this Pacific initiative has correctly been at regional and country level, but there is a growing realization that unless sub-national surveillance is strengthened, prompt effective containment will not be possible. A responsive outbreak surveillance system developed in Mpumulanga Province, South Africa, demonstrated the effectiveness of sub-national surveillance in a resource-poor setting (Durrheim et al. 2001, 2002). In this system, infection control nurses using an Outbreak Manual were the ‘eyes’ of the system reporting to a responsive central communicable disease control unit based in the Department of Health in the capital city of the province. This paper reports on the early progress in adapting and transferring this outbreak surveillance concept to a Pacific island nation.
Tuvalu is an independent Pacific country with a population of approximately 9700 residents living on nine rural islands. The total land area of 26 km2 is not arable, being coral atolls, and Tuvalu's highest point is 5 m above sea level. Tuvalu's principal agricultural products are coconuts and fish, and its remote location and attendant travel costs have resulted in less than 1000 tourists per annum. The country has a number of features that markedly increase the likelihood of communicable disease outbreaks. In addition to the limited clinical and public health infrastructure and the remoteness of the individual islands, which complicate outbreak response, waste disposal is a continuing issue of public health and environmental concern. Tuvalu is situated in a region that experiences regular destructive cyclonic activity and a major source of male employment is international sea-faring, with an ongoing risk of importation of communicable diseases not usually present in Tuvalu. The Ministry of Health recognized this vulnerability and the shortcomings of the existing passive monthly notification system, particularly under-reporting and delayed reporting, and decided to embark on a process of sub-national surveillance capacity-building.
The hub of the system is the public health physician at the Ministry of Health on Funafuti, the principal island, while the reporting units are the hospital and seven island clinics. One remote island has no nurse and only 30 residents, and was not involved in the initial phase of the surveillance system. On Funafuti the outpatient department at Princess Margaret Hospital, the only hospital in the country, was selected as a reporting unit as it provides primary health care services and all hospital admissions occur through the outpatient department. Decentralizing surveillance to the seven outer islands was considered feasible as each island has a clinic staffed by one or, on the three larger islands, two nurses; a regular electricity supply; a refrigerator, usually used for storing vaccines but available for temporarily storing specimens; and a telephone. Each island council has access to a fax machine. There are scheduled boat trips to the outer islands from Funafuti, the main island, once every 10 days, and a military launch is available for emergency transport.
After determining feasibility, consultation with senior Ministry of Health officials identified a limited number of disease syndromes considered essential for enhanced surveillance and immediate response. Inclusion was based on recognized risk, local epidemiology and potential for spread, likely public health burden, international reporting obligations and available effective means for containment. The eight syndromes chosen were: AFP (poliomyelitis); profuse watery diarrhoea (cholera); diarrhoea outbreak; dysentery outbreak; febrile disease with abdominal symptoms and headache (typhoid); febrile disease with generalized non-blistering rash (measles); febrile disease with intense headache and/or neck stiffness, with or without haemorrhagic rash (meningococcal meningitis); and outbreaks of other febrile diseases of uncertain origin. The template for this phase was the Outbreak Manual developed and used successfully in Mpumalanga Province, South Africa (Durrheim et al. 2001). The Mpumulanga manual was modified by reducing the number of syndromes from 10 to 8 by removing syndromes unlikely to be encountered in Tuvalu (plague, viral haemorrhagic fever and yellow fever), by adding a specific syndrome for typhoid, and by reducing the number of cases of diarrhoea and dysentery required to trigger an outbreak report from 10 to 5 and from 10 to 3, respectively. Typhoid was added, as Tuvalu had a history of typhoid outbreaks that had proved difficult to control.
A system of zero-reporting was agreed with the clinic nurses. If no suspected cases meeting the syndrome descriptions had occurred in a specific week, then on Friday afternoon before 2 pm a fax to this effect would be sent to the information manager at the Ministry of Health. Simple forms were prepared to facilitate zero-reporting and line-listing of cases. An Outbreak Manual was prepared to provide a systematic approach for identifying each syndrome using a clinical case definition. The manual also provided criteria for confirmation; details on standard responses (including infection control precautions) and initial therapy; discussion on the preparation of diagnostic specimens and community control strategies; and guidelines for immediate reporting (http://www.wepi.org/books/tom/). The manual was used for in-depth training of island nurses at their annual three-week training in July 2003. The nurses expressed their support for the system and indicated that they were keen to implement it.
Successful outbreak containment is the primary goal of the system and thus the true measure of its value is its performance in rapidly controlling outbreaks and limiting second generations of disease. It is, however, essential to monitor surveillance system performance on an ongoing basis (World Health Organization 2004). The proxy measure selected was the completeness and timeliness of weekly zero reporting by island nurses.
Initially, zero-reporting progress was satisfactory, with completeness of reporting rising steadily from 60% in August 2003 to 80% in October 2003. But then reporting completeness fell below 40% in December 2003 and January 2004. This happened despite a number of documented successful responses to diarrhoeal and respiratory outbreaks using the protocols in the manual, most notably a dysentery outbreak in a high-risk secondary school on Vaitupu Island. When local health staff compared the success of this response with detection and response to a previous typhoid outbreak in the same school in 2000, they attributed the success of the recent response to the new surveillance system. The differences highlighted were faster initial reporting to the Ministry and response by the central outbreak control team; a greater commitment of resources for control strategies and for infrastructure improvements after the outbreak; and increased confidence in responding because printed guidelines were available to direct diagnosis, clinical management and disease control actions.
The dissonance between surveillance indicators and successful outbreak control prompted an in-depth review of the surveillance system in February 2004. Structured system review and in-depth interviews with key national level staff and island clinic nurses were conducted. It became apparent that there had been relative neglect, during the initial consultation and training, of national staff at the Ministry of Health on Funafuti, particularly the information manager, resulting in a limited appreciation of the importance of immediate communication with clinic staff when no weekly zero-report was received, and no formal feedback to clinic staff on their performance, or that of their peers, in outbreak surveillance and response. The telephone interviews with clinic nurses on the outer islands provided valuable insights into the perceived benefits of the system: (Nurse on Niutao: ‘now know what to do...outbreaks (are) less frightening’; Nurse on Nanumea: ‘it is useful…now know the signs and response’; and Nurse on Nukufetau: ‘read Manual often…it helps me treat diarrhoea patients’). The enquiry highlighted weaknesses, particularly that nurses had to personally pay for using the island council fax machine to send zero-reports, while providing suggestions for resolving these problems [Nurse on Vaitupu: ‘will phone if (there are) cases…won't send zero-reports until clinic has own fax’; Nurse on Nanumanga: ‘would like to know what is happening on other islands’].
In response to these findings the national information manager was specifically trained and assisted in developing a simple database for collating and analyzing zero-reports and outbreak line-lists, and supported in preparing the initial editions of a monthly Tuvalu Epidemic Update, that provides feedback on recent local outbreaks, compares reporting performance by island and provides relevant additional epidemic information from the Region. Fax-phones were purchased and installed in each clinic to facilitate reporting and receipt of epidemic updates. This resulted in a resurgence in reporting with completeness exceeding 90% in all months since March 2004, except May 2004 when all nurses were on Funafuti for their annual training, which included refresher surveillance training! In addition to the obvious need to expand peripheral surveillance capacity despite limited human resources, the necessity of establishing alternative communication channels when telephone connections are disrupted and assuring access to reliable reference laboratory facilities to provide confirmatory diagnosis are remaining surveillance challenges in Tuvalu.
Island clinic nurses indicated that they were likely to become aware of all outbreaks of public health importance due to the relatively small size of the communities they serve, the absence of an alternative system of traditional health care that could influence presentation at public health facilities, and their intimate links with their communities. Thus, a more comprehensive systematic evaluation of the validity measures (sensitivity and specificity) of the surveillance system should be possible in the future as outbreaks accumulate (Buehler et al. 2004). This will add to the present descriptive evaluation of the system's design and operation, timeliness, acceptability and perceived value to users.
Global investments to improve surveillance should not exclusively focus on regional or national levels but must extend sub-nationally in all developing countries with limited capacity (Durrheim & Speare 2004). The recent Revision of the International Health Regulations (World Health Assembly 2005) requires that all member states ‘develop, strengthen and maintain, as soon as possible but no later than 5 years… the capacity to detect, assess, notify and report disease events’. This poses a major challenge to many developing countries and new simple effective surveillance models are required.
For effective containment of outbreaks, patients with infectious diseases of public health importance must be recognized and promptly reported to those responsible for prevention and control activities. This is particularly important in under-resourced regions where delays in raising the alarm may cause vulnerable communities to suffer multiple generations of disease, with unnecessary morbidity, death, panic, and loss of public health system credibility. It is not sufficient to implement what appear to be viable surveillance systems, without ongoing monitoring, and, where indicated, critical review (Jena et al. 2004). The adaptation of the Mpumalanga Outbreak Manual for Tuvalu was successful and regarded as a valuable tool by reporting nurses. Although the early progress of outbreak surveillance and response in Tuvalu is encouraging, it demonstrates the challenging nature of communicable disease surveillance in developing countries and the need for ongoing monitoring and responsive adaptation to feedback from surveillance agents based in the periphery of the health system.