Addressing malaria importation in a near-elimination setting requires an understanding of how and from where parasites are imported. Our study provides details on travel patterns between the low transmission setting of Zanzibar and mainland Tanzania, and identifies an increased risk of malaria infection in those Zanzibar residents who travel to districts in mainland Tanzania that are classified as highly malaria endemic. These findings can inform the development of tailored approaches to target the importation of malaria parasites into Zanzibar using data that can be collected routinely.
While residual local transmission persists in Zanzibar, with Anopheles arabiensis being the most abundant vector [35, 36], previous analyses of malaria data have consistently identified travel to mainland Tanzania as a risk factor for malaria infection among residents of Zanzibar [18, 21, 24, 29, 37]. Our results complement these studies by providing details about the precise travel destination and the likely origin of the malaria infections in Tanzania mainland [21, 23].
Malaria endemicity remains high in the south-east of Tanzania, including the Indian Ocean coastline south of Dar es Salaam, and in the north-west of the country [32]. Roughly two-thirds of the trips reported by our study participants from Zanzibar included visiting a district classified as highly malaria endemic according to Thawer et al. [32] while few travellers exclusively visited areas with a low malaria endemicity (which includes the centre and north of Tanzania, as well as most of Dar es Salaam). Travelers who visited highly endemic districts contributed the majority (77%) of all infections detected among travellers to mainland Tanzania. Trips to high endemicity areas were not only more frequent but also generally longer than those to other areas (including Dar es Salaam, the main port of entry to mainland Tanzania), which was expected to result in an increased chance of exposure to infective mosquito bites.
Previous studies applied mobile phone data to capture the movement of travellers between Zanzibar and mainland Tanzania and model the risk of importing parasites based on travel destinations in five large telecommunication coverage regions of the mobile phone service provider ‘Zantel’ in mainland Tanzania [22, 23]. Yet, these coverage regions are very large and comprise a variety of endemicity settings. For example, the Dar es Salaam coverage region, previously identified as the primary travel destination from Zanzibar [22], comprised the entire coastline from the Kenyan to the Mozambican border. While Dar es Salaam is where most of the travellers from Zanzibar start their journey to other destinations in mainland Tanzania, our study established that only 21% of travellers remain in Dar es Salaam without travelling onwards to other destinations, and of those who did stay, only a few were infected with malaria parasites.
Our study combined self-reported travel data that can be collected routinely in RACD household follow-up activities by ZAMEP and malaria endemicity data based on routine health information system data from mainland Tanzania [38]. The important Advantages of our approach include the higher granularity of the travel data, reflecting the heterogeneity of malaria endemicity in mainland Tanzania, and the continuous data availability through routinely implemented surveillance activities. The strong association of malaria endemicity in the travel destinations and infection among travellers demonstrated in our analysis, provides a strong rationale for prioritising those who travel to highly endemic areas with more strict measures to prevent malaria importation in Zanzibar. At the same time, it underlines the importance of reducing malaria transmission in high endemicity areas in mainland Tanzania in order to reduce parasite importation in the isles, as also suggested in a recent modelling study [24].
Considering that Dar es Salaam is the main port of entry and exit for travellers between Zanzibar and mainland Tanzania and an important commercial centre connecting regions within Tanzania and neighbouring countries, measures to reduce transmission must include preventing the invasion of more aggressive urban malaria vector species, namely An. stephensi, that could also be carried to Zanzibar via different modes of human or cargo transport [39]. In light of the high connectivity of Zanzibar and mainland Tanzania, both vector and parasite surveillance and adequate targeted response should be a priority for both the mainland Tanzania national malaria control programme (NMCP) and ZAMEP, working closely together.
Albeit available only for a sub-set of travellers recruited in the later part of the study, our data indicates that travellers were not likely to consistently use mosquito nets or other measures to prevent mosquito bites while travelling. Although there is a high mosquito net coverage in mainland Tanzania (84% use) [40], households may not have sufficient (spare) nets for temporary guests. Travellers would have to carry their own protective gear if it is not provided where they stay overnight. Risk awareness or personal preferences may further influence the use of protective measures even if they are available [41]. Our findings suggest that malaria infection and importation in Zanzibar have an association with travelling during the dry season. In Zanzibar, there is a strong relationship between seasonality and malaria transmission peaks, as a significant proportion of infections occur towards the end of the rainy seasons. The slightly lower use of mosquito nets during the dry seasons, when the temperatures are likely to be higher outside and in the room and sleeping space, may be due to discomfort when sleeping under the net, as previously reported elsewhere [42]. The observed increase in the odds of infection during the rainy season can be attributed to the timing of data collection. This is due to the fact that there were only a few observations in the rainy seasons.
The analyses in this manuscript used self-reported travel information over a period of 60 days. The risk of recall bias was minimised by limiting the analyses to broad categories of number of days in the study variables. Data collection for this study was linked to ZAMEP’s reactive case detection, that is, each data collection activity was triggered by a reported confirmed malaria case. Even though the sample included transect households (with a malaria prevalence comparable to the general population [21]), households with travellers who returned without a malaria infection might be underrepresented in this study, potentially biasing the sample towards travel to higher endemicity areas and higher malaria prevalence than what would be found in the general population. While the frequency of travel destinations might therefore not be representative of all travellers from Zanzibar, the destinations are likely to be an adequate reflection of locations from which malaria infections are imported. At the same time, our study only included five districts (excluding, for example Mjini/urban district, the district with highest number of travellers according to routine ZAMEP data, and possibly highest number of imported infections [23]), hence some of the results may not be representative of all of Zanzibar. Routine travel location data collected by ZAMEP collected through surveillance response system may be used to update our analyses of travel destinations, classified based on their level of endemicity.
Based on our findings, interventions addressing malaria importation could preferentially target individuals who travel to and from high endemicity areas in mainland Tanzania. Regular analyses of surveillance data indicating the origin of infections classified as ‘imported’ and further qualitative research may be needed to identify socio-demographic characteristics of these ‘high risk’ groups and to develop and implement effective targeted interventions. Interventions to prevent and clear malaria infections, for example malaria chemoprophylaxis, sensitization activities, or test-and-treat programmes at the port of entry, may be more feasible and cost-effective to implement in clearly defined ‘high risk’ groups rather than all incoming travellers.