Is travel prophylaxis worth while? Economic appraisal of prophylactic measures against malaria, hepatitis A, and typhoid in travellersBMJ 1994; 309 doi: https://doi.org/10.1136/bmj.309.6959.918 (Published 08 October 1994) Cite this as: BMJ 1994;309:918
- R H Behrens,
- J A Roberts
- Hospital for Tropical Diseases Travel Clinic, London NW1 0PE
- Health Services Research Unit, London School of Hygiene and Tropical Medicine, London WC1T 7HT20
- Correspondence to: Dr Behrens.
- Accepted 18 July 1994
Objectives: To estimate the costs and benefits of prophylaxis against travel acquired malaria, typhoid20fever, and hepatitis A in United Kingdom residents during 1991.
Design: Retrospective analysis of national epidemiological and economic data.
Main outcome measures: Incidence of travel associated infections in susceptible United Kingdom residents per visit; costs of prophylaxis provision from historical data; benefits to the health sector, community, and individuals in terms of avoided morbidity and mortality based on hospital and community costs of disease.
Results: The high incidence of imported malaria (0.70%) and the low costs of providing chemoprophylaxis resulted in a cost-benefit ratio of 0.19 for chloroquine and proguanil and 0.57 for a regimen containing mefloquine. Hepatitis A infection occurred in 0.05% of visits and the cost of prophylaxis invariably exceeded the benefits for immunoglobulin (cost-benefit ratio 5.8) and inactivated hepatitis A vaccine (cost- benefit ratio 15.8). Similarly, low incidence of typhoid (0.02%) and its high cost gave whole cell killed, polysaccharide Vi, and oral Ty 21a typhoid vaccines cost-benefit ratios of 18.1, 18.0, and 22.0 respectively.
Conclusions: Fewer than one third of travellers receive vaccines but the total cost of providing typhoid and hepatitis A prophylaxis of pounds sterling25.8m is significantly higher than the treatment costs to the NHS (pounds sterling 1.03m) of cases avoided by prophylaxis. Neither hepatitis A prophylaxis nor typhoid prophylaxis is cost effective, but costs of treating malaria20greatly exceed costs of chemoprophylaxis, which is therefore highly cost effective.
Public health implications
Public health implications
Travel related hepatitis A and typhoid occurred in fewer than 0.05% of visits by United Kingdom residents abroad in 1991 as compared with malaria incidence of 0.7% of visits to endemic regions
Around one third of travellers receive pretravel immunisations
The short duration of illness and low mortality from typhoid and hepatitis A give rise to minimal health costs but the vaccines have high administrative costs, which make prophylaxis cost ineffective
Malaria and its associated mortality incur appreciable health care costs, but chemoprophylaxis against malaria is easily provided and is less expensive and therefore is highly cost effective
Public subsidy for vaccinations should be reviewed and targeted
Of the 28 million British travellers in 1991, 12.6 million travelled to destinations outside North America and central Europe and 756 0001 travelled to malarious regions. Because of a perceived risk of diseases in tropical destinations, many intending travellers seek information on recommended immunisations and malaria chemoprophylaxis. To meet the demand many groups provide information about chemoprophylaxis and eminent authorities advise a range of vaccines, presuming these measures to be cost effective. Public health policy has not challenged that belief. We used economic analysis to evaluate pretravel prophylaxis in travellers. We aimed at providing an estimate of costs and benefits of various prophylactic regimens against malaria, typhoid fever, and hepatitis A. We adopted the framework of cost-benefit analysis to determine whether the prophylaxis was worth while, and within this framework we examined cost effectiveness of alternative vaccines and prophylactic regimens that could be used as part of a preventive strategy.
Methods and sources of data
Health sector costs were derived from records of a sample of patients treated in a hospital for tropical diseases. Costs to the individual were based on estimated time off work, costed according to wages and adjusted for cost of employment, as reported by the Department of Employment for 1991. Prices used were unit costs recorded in the British National Formulary No 22 (1991) for existing vaccines and in the British National Formulary No 26 (1993) for new vaccines and drugs. Benefits were computed as avoided costs of illness.
Incidence of disease in travellers
As country specific information on infectious disease was largely unknown, we estimated the incidences of hepatitis A, typhoid fever, and malaria in United Kingdom residents returned from disease endemic regions. The incidence of travel associated infections in journeys to endemic countries defined by the World Health Organisation2 was expressed as a proportion of the number of visits by non-immune and non- immunised United Kingdom residents. The incidence of travel associated hepatitis A and typhoid in United Kingdom residents was calculated from surveillance data3 by adjusting for underreporting.4, 5 Malaria cases were further adjusted for by excluding foreign visitors, immigrants, and cases in refugees (Malaria Reference Laboratory, Public Health Laboratory Service, London). Visits to countries outside eastern Europe, the European Union including its Mediterranean members, North America, Australia, and New Zealand were used as the denominator for hepatitis A and typhoid. Numbers of visits were provided by the International Passenger Survey,1 a longitudinal survey of passengers resident in the United Kingdom. The survey provides country specific departures from the United Kingdom. Numbers of overseas visits were used as the denominator and may represent repeated journeys by the same person. Ninety five per cent confidence intervals were used for the sensitivity analysis.
Provision of immunisation
Currently in the United Kingdom travel advice and vaccines are predominantly given by general practitioners as part of their service contract. The statistics unit of the Department of Health provided numbers of doses of whole cell killed monovalent typhoid vaccine and human normal immunoglobulin refunded through the NHS prescription pricing authority. These were used to estimate the numbers of travellers immunised during 1991. Prescriptions of multidose phials of typhoid vaccine suggested that around 32% of all travellers had received either a booster or a primary course of typhoid vaccine, consistent with the seroprevalence of antibodies to typhoid in departing travellers reported by Cossar et al.6 The proportion of travellers seropositive to hepatitis from published figures was 54%.6 In our analysis this was made up of 30% who were naturally immune and 24% who had received passive immunisation based on prescription data. We assumed that most of those immunised were not immune.
Several of the vaccines were not in widespread use in 1991 but it was important to include the likely costs of these newer vaccines in our estimates. In order to do this we used the 1991 prescription numbers for human normal immunoglobulin and whole cell killed monovalent typhoid vaccine and estimated the cost had oral typhoid Ty 21a, typhoid polysaccharide Vi, and hepatitis A vaccines been used instead. Private prescriptions and vaccines given in the private sector were unknown. A figure of 20% of total vaccines used was assumed with a range of 0-40% for the sensitivity analysis.
Malaria prophylaxis - The proportion of United Kingdom residents who visit malaria endemic countries and use chemoprophylaxis has been reported at about 50%.7, 8 We based the incidence of malaria on the proportion not using chemoprophylaxis. The cost of three months of chloroquine and proguanil chemoprophylaxis at 1991 retail recommended prices was used in estimating costs to the traveller.8 An estimated 30% of all prophylaxis used was prescribed,9 the cost being met by the health sector. As mefloquine was introduced in late 1991 and little used, an estimate of cost for its sole use had it been prescribed was used in the calculations.
Effectiveness of prophylaxis - All the typhoid vaccine preparations were assumed to provide similar protective efficacy of 70%.10 The duration of protection against hepatitis A of a single 5 ml does of human normal immunoglobulin was six months. A three dose course of hepatitis A vaccine protects for at least one year11 and probably for five years or more. Both regimens were assumed to protect 90% of those immunised against clinical disease. The antimalarial regimens were attributed a protective efficacy of 72% for chloroquine and proguanil and 92% for mefloquine.12
For a cost-benefit analysis it is necessary to assess the costs of the prophylaxis and compare these with the gains that are attributable to its use. To do this it is necessary to trace the costs and benefits to all those concerned - namely, the public sector, including the public health service, community services, hospital services - and costs to individuals and society in terms of loss of productive capacity and, occasionally, life. We used the cost of avoided diseases estimated from prophylaxis use in 1991 and costed prophylaxis provision to derive a cost-benefit ratio.
Health sector costs
Hospital costs - The cost of imported disease to the health sector was calculated from the mean inpatient length of stay and the costs of laboratory confirmed typhoid (8.0 days; n=58), severe hepatitis A (5.0 days; n=43), and malaria (3.5 days; n=658) at the Hospital for Tropical Diseases, London. The severity of malaria and typhoid was assumed to be the same for each bed day, as no information about the marginal cost per bed day during the illness was available. Ninety per cent of cases of hepatitis A were categorised as mild and not requiring hospital treatment and 10% as moderate or severe diseases needing hospital treatment.
General practitioner costs - The costs of giving vaccine, including the cost of an eight minute (on average) consultation with a general practitioner, were obtained from a British Medical Association survey13 and included the vaccine cost and disposable and reimbursement fees, less the patient's prescription charge when appropriate. (Neither age nor health of travellers was known. It was assumed that all travellers pay prescription charges (pounds sterling 3.75 in 1991) as they are predominantly fit and employed. This assumption effects the distribution of costs but not total costs.) Though there is a range of disease severity, we assumed that all malaria and typhoid patients saw a general practitioner twice and that patients with hepatitis A (assumed to be mainly housebound14) saw a general practitioner four times.
Costs to travellers and society
Costs to the traveller and society include lost productivity time during illness - 21 days in mild hepatitis; 95 days in severe hepatitis; 29 days for typhoid; 17.5 days in malaria. Lost income was based on average earnings plus costs of employment in 1991 for non-manual employees.15 Costs of travel to the general practitioner's surgery and to hospital for diagnosis and treatment and prescription charges and expenses of obtaining prophylaxis were also included. Children with infection require the same care as adults of working age, and we assumed a productivity loss in all cases. Some argue against using lost productivity during periods of high unemployment but, as recruitment is expensive, this is irrelevant for short term illness.
Loss of life
As some people might have avoided death by being vaccinated, our study incorporated a value of life estimate of pounds sterling 1.4m,16 a midpoint estimate from other studies. This subject is contentious but if not confronted would severely misrepresent the benefits of the programme. In addition to including a value of life in the cost-benefit analysis, we calculated the implied value of life. We used reported case fatality rates for typhoid,17 hepatitis A,18 and malaria19 in table II to provide the number of lost lives and calculated the value of these lives by the incurred costs of avoidance.
Willingness to pay
In the model we estimated benefits as avoided costs of illness. This approach does not address the fundamental question of how much people would have been willing to pay to avoid the infection. It does provide estimates of the pool of potential resources, if any, that may be available for other uses as a result of the disease prevention. Alternatively it indicates resources that could more appropriately be used elsewhere had the intervention not take place. We did not measure the value that might be placed by individuals on reduced morbidity or its impact on the quality of life. Estimates of these values are difficult but in subsequent studies we hope to explore the amount people would be willing to pay to avoid risk of death and reduction in quality of life when they are properly informed.
Comparison of vaccine protection by discounting costs
Because different vaccines afford differing durations of protection, costs of prophylactics that are administered repeatedly were discounted annually at the treasury rate of 6% to calculate the present value of prophylaxis at future administrations. To compare vaccine protection against hepatitis A for repeated journeys an arbitrary four journeys in five years was postulated.
Total visits for unprotected travellers and the numbers of cases were estimated. Reported cases were calculated by adjusting the reported data for under-reporting (table I). These cases formed the basis for estimates of expected deaths and the impact of prophylaxis on deaths. Deaths prevented (table II) form the basis for estimating the implied value of life and the social benefits of the programme.
Costs per unit of treatment drug costs - plus administration costs - presented in table III. For low cost regimens the administration expenses made up a substantial proportion of the total cost, disclosing the hidden expenditure borne by the health service and the individual for prophylaxis. Table IV includes the costs to the health sector of treating cases had the disease not been prevented. These were just over pounds sterling 3m for typhoid and nearly pounds sterling 7m for hepatitis A and were substantially lower than the costs of preventing the disease. In typhoid the largest proportion of the costs was borne by the community and health sector, and in hepatitis A the traveller carried the biggest proportion (86%) of the cost of illness (because of the loss of earnings). Malaria morbidity and mortality were costing the health sector and community (82% total costs) an estimated pounds sterling 20m annually, well in excess of the cost of prophylaxis.
Table V shows the cost effectiveness of the various types of intervention of each disease, presented as cost of intervention and treatment and the cost- benefit ratios. The number of cases prevented is the product of the incidence by doses of prophylaxis, adjusted for prophylaxis efficacy. The savings attributable to the intervention (benefit) is the avoided disease calculated from the sum of community, travellers', general practitioners', and hospital costs by the numbers of cases prevented by prophylaxis in 1991. The duration in hospital and cost of deaths are the main contributors to morbidity costs, except in hepatitis A infection, in which wage losses make the largest contribution to costs. The impact of the duration of protection was included by estimating the present value of treating cases with the assumption that several visits were made over time. Human normal immunoglobulin was threefold more cost effective for a single journey than the vaccine, but this changed with frequency of travel (figure).
With the exception of prophylaxis against malaria the costs of avoiding cases greatly exceeded the costs per case for treating the disease. This is clearly shown by the cost-benefit ratios. Except for malaria, benefits (expressed as avoided costs) did not exceed costs whatever assumptions were made (table V; ranges derived from sensitivity analysis). The sensitivity analysis was based on the confidence intervals used to provide maximum and minimum estimates and applied to all variables in the model. An adjustment of plus or minus 10% was included for the costing data.
Expenditure necessary to prevent a death (the implied value of life) from malaria was pounds sterling 412 014 with chloroquine and proguanil and pounds sterling 1 235 692 with mefloquine, which is similar to other forms of health interventions. In contrast, expenditure for preventing a death with hepatitis A vaccine was pounds sterling 187 137 215 and pounds sterling67 847 918 with oral typhoid vaccine.
We have estimated the cost effectiveness and cost-benefit ratio of prophylaxis against travel related infections. The analysis used the best available data and included several assumptions to provide cost-benefit ratios for different methods of prophylaxis, a sensitivity analysis giving a probable range of results. The data generated must be interpreted within the limitations of the model and its assumptions. The incidence of disease was a critical factor in our model: the cost-benefit ratio was very sensitive to changes in incidence. We derived our incidence rates from surveillance reports of imported infections, which rely on the inclusion of travel information. These reports were frequently of poor quality.*RF 19a* We compensated for underreporting by adjusting numbers and calculating the overall incidence of typhoid, hepatitis A, and malaria in unprotected travellers (non-immune, unimmunised, and prophylaxis failures). The model was also sensitive to changes in population seroprevalence. For a seroprevalence of hepatitis A antibodies of 7.9%, as currently found in British army recruits,20 the cost-benefit ratio would increase to 8.26 for human normal immunoglobulin and 22.0 for hepatitis A vaccine. The efficacy of drugs or vaccines and cost of illness has little impact on the cost-benefit ratio.
The assumption that each prescription reflects an immunised traveller needs to be validated. We compared the estimated 600 000 doses of human normal immunoglobulin given to travellers in 1990 reported by Tilzey et al21 with the total doses in 1991 calculated by our model (624 000). The result suggests that our assumption is realistic. The impact and cost of imported infections to local communities were not included in our analysis. Malaria poses no risk but typhoid and hepatitis A have the potential to cause disease outbreaks in the community. Salmonella typhi is predominantly transmitted from index cases who are chronic carriers, and outbreaks originating in returned travellers have not been reported.17 There are no data available to determine whether parenteral vaccination protects against asymptomatic carriage and excretion of S typhi. There seems to be an insignificant risk of imported hepatitis A causing community outbreaks, only 10 of 2016 outbreak associated cases having been linked to recently returned travellers.18
Our model allows the contribution of individual variables to the cost- benefit ratio to be determined. To provide a cost-benefit ratio of one requires the same travelling population to present with 1550 cases of hepatitis A and 3500 cases of typhoid. Reported cases in 1991 were 288 and 164 respectively. The incidence of infection in travellers depends on many factors, including the host's immune status, region of travel, and duration and degree of exposure. Our calculated incidence of hepatitis A of 0.05% was sixfold lower than the incidence in unimmunised Swiss travellers, which was based on eight imported cases,22 and suggests that the overall risk for British travellers to all endemic countries is low. But even if the higher Swiss rates were used, the costs would still exceed the benefits by a significant amount.
Methods of protection over different durations were compared by using discounted costs for multiple journeys. The inactivated hepatitis vaccine, which protects for five years or longer, was compared with repeated 5 ml doses of human normal immunoglobulin each lasting for six months, when a traveller makes three or more journeys the regimens have similar cost-benefit ratios (5.8), reflected in the figure at the crossover point of the plots. For travellers making more than three journeys the hepatitis A vaccine provides a more cost effective option. Though the incidence of malaria is 14 times higher than that of hepatitis A, cost and provision of chemoprophylaxis are much cheaper and utilisation rates higher. Hence the calculated cost-benefit ratio of 0.19 for chloroquine and proguanil and 0.57 for mefloquine prophylaxis makes both cost effective and likely to save the health service resources when given within the current system.
Implications for society and the individual
For society the estimated costs to the NHS of giving typhoid and hepatitis prophylaxis to a third of travellers was pounds sterling 25.8m compared with the estimated costs of treating (in a specialist London hospital) cases prevented by the use of vaccine of pounds sterling 1.03m. This, along with implied value of life, which is well in excess of most other lifesaving strategies, indicates that immunisation against hepatitis A and typhoid is not an efficient use of NHS resources. The cost and benefits of screening were excluded in this study because the high cost of screening and low seroprevalance of antibodies are likely only to increase the cost-benefit ratio.
For the individual, however, the perceived risk of infection and morbidity may be higher than the estimated incidence in this study. The analysis shows the proportion of illness costs borne by the travellers balanced against the cost to themselves of prescription charges and loss of earnings for a consultation. This makes the cost-benefit of immunisation with human normal immunoglobulin (cost-benefit ratio 3.4) more evenly balanced. But for whole cell killed monovalent typhoid vaccine or hepatitis A vaccine, even for the individual, costs outweigh the benefits (cost- benefit ratios 23.8 and 9.2). We recognise that there is a difference between public policy and individual preference. The value and benefits individuals would attribute to the various methods of immunisation would be linked to their perceived threat of disease and its likely impact on themselves. Some people are at greater risk or have lower thresholds about taking any health risk. Their needs could be accommodated either by a payment system or by designing a suitable protocol which would target high risk non-immune groups. These might be military personnel, overseas volunteer workers, expatriates, or elderly people, who, although they have a significantly higher morbidity and mortality from hepatitis A, have a higher seroprevalence of hepatitis A antibodies.23
With increasing numbers of travellers and marketing of more expensive vaccines, the disparity between costs and benefits will undoubtedly have risen since 1991. Data on the contribution of increasing herd immunity to disease incidence, the destination specific disease incidence, and the impact of repeated travel - especially stratified by reason and type of travel - would add to the value and sensitivity of our estimates. Immunisation has only a minor role in reducing travel illnesses, which are generally most effectively avoided by travellers adopting appropriate behaviour.24, 25 To provide substantive protection against malaria, for example, compliant and careful insect avoidance measures and drug use are necessary. Advice and information which influence and improve behaviour make the biggest impact on travel related illness and should be the priority health strategy in protecting travellers. A national policy of immunising all travellers against typhoid and hepatitis A does not seem be cost beneficial to the community, and public funding for such a policy should be critically reviewed.
We are grateful for the help and data provided by the statistics office of the Department of Health, the International Passenger Survey of the Office of Population Censuses and Surveys, and Marie Blaze of the Malaria Reference Laboratory. We thank our many colleagues for their helpful comments and discussion in preparing the manuscript.