The Challenges of Disease Risk Ascertainment Using Accessible Data Sources for Numbers of Travelers

Ronald H. Behrens MD, Bernadette Carroll BSc
DOI: 296-302 First published online: 1 September 2013


Background Accessible travel has led to a rapid growth in international tourism, particularly to developing countries. With the increase, travel‐associated morbidity and mortality has changed. Data on traveling populations are essential for policy makers to estimate infectious and noninfectious risks in travelers. Passenger flow statistics are compiled by the World Tourism Organization (WTO) and by official institutions of some countries. This study investigates sources of passenger flow statistics, methods of data collection, and compares datasets for consistency.

Methods Four national datasets of departing travelers were compared to the United Nations' World Tourism Organization (WTO) data on passenger arrivals to eight destination countries. The ratio between arrivals and departures was calculated (main destination ratio [MDR]) to estimate the proportion of direct to indirect traveler arrivals.

Results With few exceptions, arrival data exceeded that of departure data for all destinations. India is a primary destination for Australian residents where arrival and departure figures were similar (MDR 1.1), while visits to Cambodia and Turkey, with 3.6‐ and 3.8‐fold higher arrivals, respectively, are part of multidestination trips. For UK residents, arrivals exceeded departures for all destinations except India where the reverse was true (MDR 0.8). A close correlation between arrivals and departures was noted for visits to South Africa while arrivals to Singapore and Cambodia were 7‐ and 10‐fold higher, respectively. Arrivals by Finnish residents to destination countries were 1.4‐ to 1.6‐fold higher than departures and 2.2‐fold higher for Canadians visiting China.

Conclusions Different methodologies used to capture arrival and departure statistics result in different estimations of traveler numbers. Data from a single source does not provide a comprehensive picture of most tourism itineraries. Inbound statistics give a more accurate reflection of the total visits made by travelers from a source country.

The tourist industry is one of the largest and fastest growing economic sectors of the world and has a wide‐ranging impact on receiving countries and their economies. Reliable statistics are necessary to define a denominator for travelers' health risks and measure morbidity and mortality among traveling populations. The World Tourism Organization (WTO), a specialized agency of the United Nations collates data from official national institutions, territories, and bodies including national tourist organizations and the International Monetary Fund.1 Other sources include national datasets collected as part of the balance of payments accounts and data collected by commercial organizations. The purpose of this study is to compare, by methodology and findings, selected accessible sources of passenger flow statistics. WTO data (passenger arrival data or inbound flow) was linked to outbound data, collected nationally, from four countries, using various methods of collection. The strengths and weaknesses of the methods are examined and the source and methodology most likely to provide a representative travelers' health dataset are discussed.

Study Methods

We compared WTO passenger arrivals data in eight destination countries, six Asian (Cambodia, China, India, Singapore, Thailand, and Vietnam), one African (South Africa), and one European (Turkey) across five consecutive years (2005–2009) with outbound (departing) nationally collected data from four source countries, Australia, the United Kingdom, Canada, and Finland. We approached all countries by e‐mail and requested detailed information on methods and sources used for collecting the dataset. Source countries were selected where details of travel by nationals or residents of these countries were available, without cost, from official websites. All the destination countries (except Singapore) are UNWTO member states.2 We calculated the ratio between WTO country arrivals and source country departures to derive a main destination ratio (MDR)3 in order to estimate the number of direct (single destination) and indirect (multidestination) travels to the destination countries.

Arrival Statistics (Inbound Flow)

World Tourism Organization

WTO systematically gathers and records tourism statistics from countries and territories around the world on inbound, outbound, and domestic tourism. Since the 1980s, a variety of data on World Tourism have been collected from over 200 countries and territories.1 The data are collated from many sources including: border records (police, immigration, traffic counts, and other types of controls), border surveys, and registrations in accommodation establishments.4 To standardize the process, the WTO created a common reference framework to measure flows of inbound and outbound visitors.5,6 Standard definitions for compiling tourism statistics are shown in Table 1.

View this table:
Table 1

World Tourism Organization recommended definitions for the compilation of tourism statistics

UNWTO definitions relating to tourism
VisitorA visitor is a traveler taking a trip to a main destination outside his/her usual environment for less than a year, for any main purpose (business, leisure or other personal purpose) other than to be employed by a resident entity in the country or place visited.
TripA trip refers to the travel by a person from the time of departure from his/her usual residence until he/she returns: it thus refers to a round trip. A trip is made up of visits to different places.
Travel/travelerTravel refers to the activity of travelers. A traveler is someone who moves between different geographic locations, for any purpose and any duration.
Tourism tripTrips taken by visitors are tourism trips.
Tourist (or overnight visitor)A visitor (domestic, inbound, or outbound) is classified as a tourist (or overnight visitor), if his/her trip includes an overnight stay, or as a same‐day visitor (or excursionist) otherwise.
Usual environmentThe usual environment of an individual, a key concept in tourism, is defined as the geographical area (though not necessarily a contiguous one) within which an individual conducts his/her regular life routines.
Country of residence and nationalityThe concept of “country of residence” of a traveler is different from that of his/her nationality or citizenship. While nationality is indicated in the traveler's passport (or other identification document), the country of residence has to be determined by means of a question (usually the indication of the current home address), although this might not be sufficient. It is recommended that travelers (and visitors) be classified on the basis of their country of residence.
VisitA trip is made up of visits to different places. The term “tourism visit” refers to a stay in a place visited during a tourism trip.

The WTO statistics in this study are extracted from international inbound traveler data,4 which details arrivals to the destination countries. The data are collected as either arrivals by country of residence or nationality.

Destination Countries

Arrivals in Cambodia

Inbound travel to Cambodia is compiled by the Ministry of Tourism, Phnom Penh,7 and compiled in accordance with the WTO tourism framework. Arrival cards from 21 international border checks in Cambodia (personal communication) are used as source data.

Arrivals in China

Data on the number of overseas arrivals to China by country/region of residence are collected by the Ministry of Public Security and State Tourism Administration and are available from the China Statistical Yearbook,8 with details of the methodology on the National Bureau of Statistics of China website.9

Arrivals in India

Data on inbound arrivals to India are compiled by the Ministry of Tourism.10 The data are generated from disembarkation cards at 72 ports of entry into India, supplemented with information from the bureau of immigration.

Arrivals in Singapore

Arrivals to Singapore are compiled from disembarkation/embarkation forms completed by all passengers arriving in Singapore. Details are available in the Yearbook of Statistics 2012 and Annual Tourism Statistics Reports.11

Arrivals in Thailand

Data on inbound visits to Thailand are compiled by the Office of Tourism Development, Ministry of Tourism and Sports.12 No details of their data collection methodology are available.

Arrivals in Vietnam

Data on international visitors to Vietnam are compiled by the Vietnam National Administration of Tourism, Ministry of Culture, Sports, and Tourism.13 No details of their data collection methods are available.

Arrivals in South Africa

Information on foreign arrivals to South Africa are compiled from data collected by immigration officers at all land, air, and sea entry ports from all residents and foreigners arriving in or departing from South Africa. Data on inbound travel, with details of methodology and definitions, are available from Statistics South Africa.14

Arrivals in Turkey

Data on arrivals of tourists by nationality are derived from border statistics on inbound travel by Turkish citizens and available from the Ministry of Tourism and Culture website.15

Departure Statistics (Outbound Flow)

Source countries

Departures from Australia

Inbound and outbound travel to, and from, Australia is collated by the Australian Bureau of Statistics (ABS), Australia's National Statistical Office. The data are from passenger cards, supplemented with information from visa applications. Data on short‐term visits of Australian residents were extracted from the ABS website.16 Overseas departure statistics reflects departure of persons through Australian air or sea ports and only details of the country where most time will be spent abroad are collected. Further details of the methodology are available on the ABS website.17

Departures from the UK (International Passenger Survey)

The International Passenger Survey (IPS) is a large multipurpose survey that collects information from passengers entering and leaving the UK. The survey covers around 95% of all passengers entering or departing the UK and a sample of around 0.2% of all travelers are randomly selected for interview. Details of the IPS methodology, coverage, and availability of datasets are published by the Office for National Statistics.18

Departures from Canada

Statistics Canada provides details of travel to and from Canada as part of the International Travel Survey (ITS).19 Data are compiled using two methods: the “Frontier Counts Survey,” which is collected using the customs declaration cards, and through questionnaire surveys (the mail‐back questionnaire survey). The mail‐back questionnaires are distributed to Canadian residents on re‐entry to Canada by Canadian Border Services Agency. Details on all places visited are collected, even if there was no overnight stay.20

Departures from Finland

Information on international travel by residents of Finland (The Finnish Travel Survey) is compiled through two separate, sample‐based telephone surveys. One is the travel survey conducted monthly that collects data from 2,350 interviews, and the other is an annual survey that collects data from 2,150 to 2,400 individuals who traveled during the past year. Data and methodology are available from Statistics Finland.21,22


Comparison of destination country arrivals with source country departures

Arrival and departure data of Australian residents to destination countries

The most frequently visited countries by Australians were Singapore (3.7 million), Thailand (2.9 million), and mainland China (2.8 million). Arrivals ranged from 5.1% to 280% higher than departures over the 5 years. The MDR was greater than 1 (indirect arrivals greater than direct) for all countries across the 5 years with the largest 3.8 for visits to Turkey, 3.6 for Cambodia, and 3.5 for Singapore (see Table 2). Closer correlations were noted for India where the 5‐year MDR was 1.1, Vietnam (1.4), and South Africa (1.6). Figure 1 shows the annual MDR variance between the WTO arrival data and the ABS data.

View this table:
Table 2

Comparison of WTO arrival data and source country departure data between 2005 and 2009 reported by national authorities and the ratio of the two sources expressed as the multidestination ratio. Annual data available in Appendices S1–S4

Australian residentsUK residentsFinnish residentsCanadian residents/nationals
WTO ArrivalsABS DeparturesMDRWTO ArrivalsUK IPS DeparturesMDRWTO ArrivalsFinland DeparturesMDRWTO ArrivalsCanadian DeparturesMDR
South Africa*454,988293,4001.62,427,1762,233,0311.146,072251,226
  • The MDR reflects the average across the 5 years.

  • * Nonresident tourists at national borders by country of residence.

  • Nonresident visitors at national borders by country of residence.

  • Nonresident tourists at national borders by nationality.

  • § Nonresident visitors at national borders by nationality.

Figure 1

Plots the annual MDR of two countries, the UK and Australia, describing the annual variation of ratio of direct to indirect (MDR) passenger numbers to eight destination countries.

Arrival and departure data of UK residents to destination countries

Turkey was the most frequently visited of the eight destination countries, with 9.2 million visits over the study. India and Thailand received 3.7 million visits, China received 2.7 million, South Africa and Singapore 2.4 million, and Vietnam and Cambodia each received less than half a million visits. Indirect arrivals exceeded direct for all countries in all years except for South Africa 2006 and for India throughout the study. Arrivals to India were lower than departures from the UK (3.7 million and 4.5 million, respectively) with an MDR of 0.8 over the 5 years. Cambodia had nearly 10 times the number of indirect visits (MDR 9.9), Singapore (MDR 7.0), and Vietnam (MDR 3.1) similarly had more indirect arrivals. South Africa (MDR 1.1) and Turkey (MDR 1.2) showed similar numbers of indirect and direct arrivals of UK residents (Table 2).

Arrival and departure data of Canadians to destination countries

China, India, and Thailand were the most popular countries for Canadian nationals/residents with 2,591,700, 988,857, and 725,608 arrivals, respectively. The remaining countries received fewer than 400,000 arrivals.

Out of the eight destination countries, China was the only one where data were available from the Statistics Canada website. The sample sizes for the remaining countries were too small to gain reliable statistics. The arrivals to China were 119% higher than departures (5‐year MDR of 2.2), see Table 2.

Arrival and departure data of Finnish residents to destination countries

The most popular destinations were Thailand with 644,950 arrivals, Turkey (471,245), and China (335,186). Arrivals were 32% to 63% higher than departures from Finland. No data were available on direct travel to Cambodia, Vietnam, South Africa, and Singapore. For the remaining four countries, indirect travel exceeded direct travel: China (MDR 1.6), India, Thailand, and Turkey (1.4); see Table 2.


Ascertainment of health threats

Over the past six decades the WTO has described a 38‐fold increase in international tourist arrivals, from 25 million in 1950 to 940 million in 2010, and projected international arrivals to reach 1.6 billion by 2020.23 Many new tourist destinations are in developing and emerging countries with a steadily increasing proportion of tourists: 31% in 1990 to 47% in 2010.23 The burden of morbidity in this population is significant as approximately 8% of travelers require medical care,24 around 1% require hospital treatment. To understand the risk of morbidity and mortality the frequency of health events in travelers needs to be adjusted by the number of travelers exposed. This relies on both an accurate numerator (ie, the number of travelers with illness or injury) and denominator (ie, the number and characteristics of travelers visiting the destination). The numerator relies on health surveillance and reporting systems in host countries. For many infectious diseases this is mandatory and, despite under‐reporting, is considered a representative reflection of imported disease.

The use of denominator data enables the calculation of incidence of morbidity, providing travel health policy makers with the most accurate picture of health threats. For example, where denominator data have been used to assess the risk of malaria among travelers,25–28 a reduction in the incidence of malaria has been shown, leading to policy change by some countries. Death rates in travelers have used denominator estimates which, in Finnish travelers, showed all mortality rate estimates of 331/100,000 person years exposed, of which the injury mortality rate was 73.5/100,000 person years exposed. This rate estimate was calculated based on the Finnish Travel Survey denominator data.29 Similar rates in US travelers globally (75/100,000 person years exposed) have been reported using other denominator data.30 The MDR for the Finnish Survey data from this study was approximately 1.5 and therefore the denominator estimate may be 50% higher than that used by Lunetta. This would lead to an overestimate of mortality by up to 50%, if Finnish travelers made multidestination journeys. The fact that most travelers undertake multidestination travel in Southeast Asia means that identifying the country or region where an individual acquired an infectious disease from this region becomes very difficult, unless there is a precise incubation period.

Consistency of direct and indirect datasets

To examine the consistency between direct and indirect travel data, we compared WTO arrival statistics with nationally collected departure data in four countries. We found that the WTO arrival statistics exceeded national data from all four source countries for each of the destination countries, with few exceptions. Arrival and departure statistics are based on quite different data‐capture methods. For arrival data, each arrival corresponds to one inbound trip, if a person visits several countries during the course of a single trip; the arrival in each country is recorded separately, whereas the WTO recommends that for departure data only the main destination of a trip is reported.5 Departure data do not therefore give an accurate picture of all the countries visited by an individual and do not account for multidestination trips. Speculative estimates can be made on the proportion of travelers for whom a given country might be a secondary destination. For example, in Table 2, 1,661,700 Australian residents departed for Thailand (2005–2009) and 2,938,183 were recorded arriving in Thailand. The difference of 1.27 million may represent Australian travelers for whom Thailand was a secondary destination and therefore not captured on the Australian departure card. A proportion of arrivals will also reflect backtracking, where the visitor visits the same country two or more times during a single trip. Data from Leiper's study show Singapore as a secondary or a stopover point in transit for the majority of Australian travelers, supporting the data in this study where there were 3.5 times more arrivals than departures from Australian residents. Leiper revealed that Hong Kong and Singapore were only the main destinations for Australians' visits in 24% to 43%, and 20% to 27%, respectively, and the remainder arrived as secondary destinations.

Visits to Singapore and Cambodia are not the first (direct) destinations for most UK residents (MDR 7.0 for Singapore, 9.9 for Cambodia) and travelers will have arrived here as part of a multidestination journey. This study highlights that travel to destinations in Asia, with few exceptions, are part of a multidestination journey. The exceptions are UK residents traveling to India, where departures from the UK were greater than arrivals throughout the study period, and Australian residents traveling to India where arrival and departure figures were similar. Part of the explanation for the difference in inbound and outbound figures from the UK to India may be that India reports arrival statistics by nationality rather than country of residence and nationals resident in the UK returning to visit friends and relatives (VFR) are excluded from the arrival data.31 In 2007, almost half a million (455,950) VFR trips were made from the UK to India.32

UK travelers to Turkey and South Africa have an MDR of approximately 1 and therefore these would be a main destination for most travelers. Unsurprisingly the MDR for Australian nationals visiting Turkey (3.8) is part of a multicountry visit but is a main destination for most Australians visiting South Africa (MDR 1.6).

Countries vary in their approach to the collection of departure data. Some are based on departure cards (eg, the ABS), others use sample surveys (eg, the UK IPS and Statistics Finland), or a combination of both (eg, Statistics Canada). Sample surveys are a cost‐effective and practical way of collecting data from a large section of the population, but for countries where fewer visits are made the sample sizes are small resulting in higher sampling errors and large confidence intervals. Where self‐completed questionnaires are used to collect data, the often low return rate may adversely influence the results and recall may be a problem where the period between travel and data collection is prolonged.

Data on regional travel and other characteristics of travel

Many of the diseases that pose a threat to travelers will have a defined area of risk. For example, the World Health Organization provides detailed regional‐ and location‐specific information on the risk of malaria by country.33 Data on regional travel and ideally travel to popular tourist locations are needed. Of equal importance to the travel health advisor is a breakdown of the characteristics of international travelers, that is, their reason for traveling, their age, and duration of stay, as these factors have a bearing on the traveler's risks and patterns of morbidity. High‐risk groups, for example, immigrants returning to VFR and older travelers, have been identified through clinical and epidemiological studies and with better denominator data more precise risk information and prevention strategies can be formulated. Although national bodies usually do provide a breakdown by some of these variables, the data are not often available or only available at a cost. The WTO does not publish such data in their yearbook of tourism statistics.


Precise statistics on the volume and flow of international travelers are essential to inform health policy. The main sources of quantitative data on tourism have been from records of arrivals at frontiers or from accommodation statistics. A more recent approach to the collection of data has been sample surveys of travelers, conducted at airports, by telephone, or by postal questionnaire.

In this study, we compared departure data from four source countries, using different methods of data collection, with arrival data in eight destination countries, between 2005 and 2009. We found that, with few exceptions, arrival data exceeded that of departure data across the 5 years. The MDRs (direct versus indirect arrivals) in most Asian countries show that visits by residents from the source countries are not direct arrivals but are part of a multicountry journey. Turkey and South Africa on the other hand appear predominantly to be direct destinations for most UK residents. Multidestination travel is the main factor in the difference observed between the arrival and departure data, but as none of the datasets make reference to multidestination travel it is not possible to judge the extent of this.

The use of data available through the web and freely accessible is a limitation in our design, and there may be other datasets which may be available but less accessible.

Accurate assessments of health risks are only possible where sufficient and credible travel denominator data are available. Arrival and departure data, collected primarily for the purpose of measuring expenditure among incoming and outgoing passengers, provide only a crude indication of the significance of flow of travelers to a destination; other factors are important when defining travel‐related health threats: location of travel within a country, the purpose of travel, types of leisure activities, modes of transport, the type of accommodation, and length of stay. Such data are limited.

Declaration of Interests

R. H. B. is supported by UCL Hospitals Comprehensive Biomedical Research Centre Infection Theme. Otherwise the authors state that they have no conflicts of interest to declare.


We would like to thank the staff of the IPS who have over many years provided detailed data from their annual survey for our research. We acknowledge the detailed information provided on request by Statistics Finland, Statistics Canada, and the ABS. We thank Calvin Jones, Cardiff Business School, for his advice.


View Abstract