HPM, specifically in low transmission and elimination settings, is important to estimate for the success of control and elimination programmes. As the failure of previous control efforts shows, neglecting HPM from high to low transmission areas, can lead to re-emergence in areas where low transmission had been previously achieved but receptivity to infection remained high [27
]. Countries aiming for elimination have, therefore, been advised by the WHO to carry out elimination feasibility assessments [93
], which encourages the use of evidence-based methods to estimate infection importation risk. Among the various types of HPM illustrated in Figure , those with certain spatial and temporal characteristics and specific demographic and socioeconomic characteristics may be more likely to travel and import infections. As Figure shows for Kenya, males between the ages of 15 and 24
years are more likely to migrate (assessed according to current and previous location of residence) than children in the same population. According to previous P. falciparum
malaria-relevant HPM analyses done in Zanzibar [15
], susceptible residents travelling from the low transmission environment of Zanzibar to higher transmission areas (mostly in mainland Tanzania) and returning with infections may be more likely to result in onward transmission in Zanzibar compared to Tanzanian residents travelling to Zanzibar, primarily due to the length of stay in high transmission location and duration of infectious period spent in Zanzibar [23
]. Plasmodium falciparum
-infected individuals travelling to low transmission, high receptivity areas pose a larger concern for elimination programmes than infected travellers moving to high transmission, high receptivity areas or low transmission, low receptivity areas. Furthermore, as national surveillance systems may detect symptomatic imported cases, asymptomatic parasite carriers and infected non-healthcare seekers are likely to go undetected unless identified through active case-detection or individually screened at entry points [94
]. Large-scale screening can be a significant expense and it would be financially difficult for most malaria elimination regions to sustain [20
]. However, identifying, testing and treating high-risk traveller groups that could potentially be targeted for specific preventive control measures, such as sugarcane plantation migrant workers in Swaziland from Mozambique [95
], may be more cost-effective. Quantitative understanding of the details of HPM patterns is useful for assessing elimination feasibility and feeding into models that can assess the operational and financial burdens of different strategies. Making precise HPM quantifications to obtain malaria-relevant details, especially where data is scarce, poses a challenge to feasibility assessment projects. However there are plenty of dispersed datasets (Tables
and ), which if carefully examined, can provide a starting point for further malaria-relevant HPM investigations (Figure ).
The datasets discussed here illustrate the HPM types that can be quantified (Table ) and where gaps may exist. Figure provides an indication of the likely origins of cross-border migrants, indicating the need for more detailed quantitative information for short-term cross-border movement, as migrants likely to visit their origin countries in the future. The existing data do however cover a large variety of HPM patterns. In general, census data provides useful information on long term HPM, indications of family ties that drive short-term visits, and demographic characteristics for national populations. Census data can also be used to assess population composition (Figure ), useful when devising infection detection and control methods, as risk of infection differ between different demographic groups [89
]. Household surveys provide data to address both long-term and short-term movements for nationwide samples, providing details on types of travel, such as family visits and vacations. Other surveys that focus on smaller geographic areas and specific sub-populations, for example border point surveys, may provide even more detailed HPM data than nationwide surveys, such as reason for travel and mode of transport used [96
]. Routine HPM data remain rare. High spatial and temporal resolution HPM data collection methods are generally expensive, but may be effectively used to capture routine HPM.
Considering the various constraints on individual datasets, using multiple, complementary datasets (Tables
and ) allows for a more detailed understanding of HPM. However, some data gaps will likely remain unfilled. For example, duration of stay in high transmission areas is amongst the more important malaria-relevant HPM metrics, but is rarely available from census and survey data. Similarly, other gaps in survey and census data include travellers' use of prophylaxis, place of stay upon travel and activities engaged in upon travel. Additionally, some types of HPM are more readily quantifiable from the data available compared to others. Routine international and cross-border HPM are difficult to quantify from existing data. Some HPM data, such as previous trips records from household survey data (Table ), may provide an indication of HPM seasonality (using time of most recent move records). However, quantifying precise seasonal inferences is a challenge. Various surveys record the number of trips made in last 12
months or time spent away in the last 12
months (Table ), however they do not give an indication of locations visited, providing an incomplete platform for assessing malaria-relevant HPM. Some malaria-relevant HPM may also go unrecorded, for example the large influxes of refugees, internally displaced people and illegal immigrants who do not disclose cross-border relocation [97
]. Finally, datasets differ from place to place and household surveys done in one country may not adequately capture relevant movements elsewhere, or be undertaken with the same set of questions. Adding questions to existing surveys, on such aspects as place and duration of stay in visited locations, travellers’ use of bednets and prophylaxis, malaria episodes and activities during travel that may increase risk of infection, e.g. farming, would improve the utility of survey HPM data in estimating infection acquistion. Moreover, standardizing such survey questions between different locations would allow for more rigorous between-country comparisons. Recording travel patterns over time using longitudinal study designs, may also enable seasonal HPM inferences from survey data.
Recently, mobile phone usage data have been used to capture nationally comprehensive, high spatial-temporal resolution, individual-level data on within country HPM and link it to disease data [98
]. However, although individual call volumes could be used as a proxy for the socio-economic status of phone-users, mobile phone usage data do not directly capture demographic and socio-economic descriptions. The potential exists though to combine such data with demographic descriptions available from surveys, providing valuable detail on the demographics of HPM. Additionally, high resolution HPM information from mobile phone usage could potentially be used to parameterize HPM models. For example, directional HPM data and distance estimates (e.g. road distances and approximate travel time obtained using road networks in the GIS framework) between locations may be used to parameterize gravity-like models [99
], and demographic stratifications of directional HPM may then be used to develop more detailed gravity models. The high resolution HPM data, such as travellers’ duration of stay in high transmission locations, may be combined with existing transmission models [36
], prevalence maps [32
] and population distribution maps [86
] to quantify imported P. falciparum
infections (Figure ). Modelling may then be used as tool to overcome uncertainties where HPM data does not exist [88
] and inform policy makers, within the bounds of uncertainty, on how to mostly effectively invest in control or elimination plans. Furthermore, beyond the survey data that exists, new approaches to collect detailed malaria-relevant HPM data may be explored. For example, incorporating additional questions on travel history in existing data collection systems [100
], as described above, would provide a single source for both individual infection risk and travel characteristics of persons enumerated. With HPM data being sparse and uncertain, alternative data sources, such as temporal sequences of satellite imagery of night time lights, may be used to assess the changing population densities of settlements through variations in illumination from fires and electric lighting, where large-scale seasonal migrations occur and compliment other HPM studies at a settlement resolution [101
]. Finally, as receptivity is critical for assessing the patterns of onward transmission instigated by imported infections, compiling historical Pf
PR data of relevance for receptivity mapping would aid future predictions of outbreaks and control needs, providing that factors such as urbanization and land use change that can permanently alter receptivity are accounted for. Projects such as The Human Mobility Mapping project [103
] aims to provide open access to HPM databases and modelling frameworks through which malaria-relevant movement parameters can be quantified.