Secondary Logo

Journal Logo

Section II: Economic Impacts

Macroeconomic and household-level impacts of HIV/AIDS in Botswana

Jefferis, Keitha; Kinghorn, Anthonyb; Siphambe, Happyc; Thurlow, Jamesd

Author Information
doi: 10.1097/01.aids.0000327631.08093.66



The detrimental impact of HIV/AIDS on public health in Africa is well documented, and most governments now recognize the need to mitigate its effects on human welfare. The macroeconomic impact of the epidemic is, however, not clear cut. Although less overt than health effects, macroeconomic impacts can become important. For example, if HIV/AIDS lowers economic growth then welfare losses will extend beyond infected individuals and their households. Fortunately, in most countries the epidemic has not been severe enough to have discernable macroeconomic effects. These impacts could, however, be substantial in southern Africa, where HIV prevalence is particularly high. For example, approximately one in every four adults in Botswana is now infected, and life expectancy fell from a peak of 62 years to 46 years in 2001. In such a case, indirect macroeconomic impacts may become important, although the magnitude and direction of these impacts depend on many factors and are difficult to predict a priori. Although gross domestic product (GDP) growth is likely to be reduced, per capita incomes and unemployment could rise or fall. Much of the early analysis of the economic impact of HIV/AIDS concluded that GDP per capita was likely to rise. More recent modelling concludes that indirect macroeconomic impacts may significantly worsen the already severe direct economic impacts felt by the infected population, while also reducing incomes among the uninfected. In such cases, macroeconomic impacts would further underscore the need to mitigate the effects of HIV/AIDS [1,2].

Most countries in southern Africa have developed national responses to the epidemic. For example, Botswana's treatment strategy is now the most extensive programme of public antiretroviral therapy (ART) provision in Africa (relative to need). This programme has complemented earlier initiatives by private medical insurers, and by 2006 a majority of individuals with clinical need were thought to be receiving treatment. Although ART rollout undoubtedly has a positive effect from a social and humanitarian perspective, however, it also raises significant fiscal concerns for the government. For example, whereas demographic projections indicate that AIDS mortality has already peaked in Botswana, the number of ART recipients is expected to double over the next decade, and illness and deaths from HIV/AIDS will continue to climb. Therefore, not only will the epidemic itself have a significant impact on Botswana's economy, but so too will the cost of providing treatment. Macroeconomic assessments must therefore account for both the impact of the epidemic and the cost of providing treatment.

In this paper we summarize our findings from a study that assessed the impact of HIV/AIDS on growth and poverty in Botswana, and on the structure and sustainability of the country's public finances [3]. Our study also considers the impact of treatment, including ART, thereby offering a quantitative basis for the decisions facing policy makers. Using a combination of economic, financial and demographic models, we provide a comprehensive assessment of both macroeconomic and household-level impacts. We find that the seemingly prohibitive cost of providing treatment is deceptive once the cost of not providing treatment is calculated, and recommend that countries in which the prevalence of HIV is particularly high perform similar macroeconomic assessments.


Impact channels

Three broad impact channels are captured in this study: macroeconomic, household, and fiscal. The effects of AIDS on key macroeconomic variables (GDP, average incomes, savings and investment, employment and wages) can broadly be divided into those arising from morbidity (illness) and mortality (death). Higher morbidity lowers productivity (time off from work or poorer performance at work), increases health-related expenditures (healthcare and sick pay), reduces savings (by lowering incomes and raising expenditures), and limits investment (as a result of lower profits, economic uncertainty, or lower savings). Rising mortality decreases the size of the population and labour force, causes a loss of skills, and changes the age structure of the workforce (which determines on-the-job experience and productivity). Therefore, HIV/AIDS affects national production or GDP through numerous channels. As each of them is likely to yield negative results, it is expected that HIV/AIDS will reduce a country's GDP. The effect on GDP per capita is indeterminate, however, depending on whether the loss in GDP outweighs the decline in the population. Similarly with unemployment, a reduction in (the growth rate of) labour supply could lead to lower unemployment, especially if labour market rigidities prevent markets from clearing, perhaps because of statutory minimum wages or trade union activity. Workers may also move from the (low wage/productivity) informal sector to better-paid employment in the formal sector. The labour market impact depends on whether supply effects, which tend to push up wages and reduce unemployment, outweigh demand effects from reduced investment and economic growth, which would have the opposite impact.

Household-level impacts may differ from macroeconomic effects. Not all households will be affected by AIDS in the same way, most obviously depending on whether a household includes an HIV-positive member. Households with an infected member may be directly affected through reduced income, increased dependency ratios (i.e. the number of non-working members supported by income-earning members), and additional expenditure on healthcare and funerals. Whereas some of these effects are permanent and others are transient, all tend to reduce disposable household income, pushing some households below the poverty line and deepening poverty for others. Overall, the direct effects of HIV/AIDS are expected to worsen poverty, although the impact on a household's incomes is likely to be worse in the short rather than the long term because some effects are temporary. A number of indirect effects also stem from the macroeconomic impacts described above. Employment levels and wages are likely to be affected, thus touching all households, not just those with infected members. Some households, such as those with skilled working members, may benefit from rising wages or from improved employment opportunities as labour markets tighten. The poverty impact will also depend on the incidence of HIV/AIDS across the income distribution. If those directly affected are predominantly from wealthier households (above the poverty datum line), and indirect effects provide income and employment opportunities for lower income households, poverty may not increase. Therefore, we cannot make a priori assumptions about the overall effect of HIV/AIDS on individual households.

The fiscal impact of HIV/AIDS depends on the scale and nature of the epidemic. Government spending is expected to rise as a result of the additional costs of AIDS treatment, care, and prevention measures. The impact on health budgets will depend on a number of factors, including: (1) the scale of need for various types of care; (2) decisions about making ART and other treatments available and the nature of the adopted service models and treatment protocols; (3) the degree to which care for AIDS patients is split between hospital-based and other types of care; (4) the ability of the public health service to manage its functions and spend available budgets; and (5) the future availability of new treatments or preventative technologies. Social welfare spending may increase to fund orphan support programmes, or alternatively fall because premature deaths reduce expenditures for old-age pensions. Conversely, impacts on government revenues are not likely to be as direct as on expenditures. For example, tax revenues in Botswana come mainly from the minerals sector, in which production is determined by mineral rather than labour resources. Furthermore, although personal taxes will be affected by population growth or decline, they account for only a small proportion of total revenues. Therefore, the fiscal impact of HIV/AIDS will depend on the demographic changes resulting from the epidemic, as well as on the nature and success of the treatment programme.

Economic models

We integrated demographic and financial projections within economic models in order to capture the three impact channels described above. At the core of the study are two types of economic models. The first is a macroeconomic growth model, which translates changes in labour supply and productivity into economic growth. The Botswana growth model distinguishes between skilled/unskilled and formal/informal labour, and captures the impact of HIV/AIDS on GDP through changes in labour supply and productivity. This class of model has been widely used in the literature because of its tractability and transparency (see studies on Tanzania [4], Malawi [5] and Botswana [6,7], and for a description of the model used in this study, Jefferis et al.[3]). The second type of model is a computable general equilibrium (CGE) and microsimulation model. Unlike aggregate growth models, CGE models are more disaggregated and take into account labour, capital and commodity markets [8]. Unlike previous studies, the CGE model used in this study distinguishes between skilled/unskilled and male/female labour, and disaggregates production and employment across sectors and regions, including metropolitan centres, small towns, and rural areas (see Thurlow [9]). The CGE model therefore captures the impact of HIV/AIDS on growth and employment at the sectoral and regional levels. As with the macroeconomic growth model, the CGE model draws on the population projections from the demographic model. The model also captures changes in government expenditures, projected by the financial projections model, and endogenously determines revenues, savings and investment. Finally, the CGE model separates households according to location/region, sex of household head, and income levels. These households are linked to a survey-based microsimulation module, which captures impacts on poverty [10]. Both types of models are calibrated to Botswana economic data for the period 2001–2003 and run forward until 2021 in line with the demographic and financial projections [11]. The CGE model also incorporates differences in HIV prevalence rates by age, sex and region [12]. Based on the demographic projections, three simulations were run capturing the current With-AIDS scenario, the hypothetical Without-AIDS scenario, and a Treatment scenario that combined Botswana's current situation with its government's treatment programme (Fig. 1).

Fig. 1
Fig. 1:
Projected HIV and AIDS mortality with and without treatment. ‘With-AIDS’ is the scenario with HIV/AIDS but without antiretroviral therapy (ART); and ‘Treatment’ is the scenario with HIV/AIDS and ART. It is assumes that ART uptake will reach approximately 90% of those people in need. Projections from the Centre for Actuarial Research, University of Cape Town.

Both models incorporate changes in labour supply, investment and productivity growth. Modelling investment in Botswana is complicated by the fact that there is a surplus of savings over investment, and thus capital outflows from the country. Investment is treated differently in the two models: in the CGE model it is savings driven, whereas in the aggregate growth model the investment rate is driven by deteriorations in the investment climate as a result of increased risks, higher costs and reduced profitability caused by HIV/AIDS. There are few empirical estimates of the productivity impacts of HIV/AIDS (see Fox et al.[13]), and parameters in the two models here are adapted from those used in previous modelling exercises (for details see Jefferis et al.[3] and Thurlow [9]).

Underlying the economic models are a set of financial projections corresponding to the three scenarios described above (Fig. 2). We first estimated levels and trends in utilization for various health and welfare services from demographic projections and existing utilization data. Utilization estimates were then combined with cost data to project expenditures. Direct estimates of unit costs for ART and orphan care are available for Botswana, but step-down costing was required for most health services. For prevention and home-based care, we based estimates on historical expenditures, current treatment programme targets and policy directives. In certain cases ‘realistic scenarios’ were developed in consultation with key stakeholders, taking into account real trends and expert opinion. Financial estimates were adjusted to reflect private sector expenditure as well as donor funding of certain health services.

Fig. 2
Fig. 2:
Simulation results for gross domestic product growth rates under the three scenarios. ‘With-AIDS’ is the scenario with HIV/AIDS but without antiretroviral therapy (ART); ‘Without-AIDS’ is the hypothetical scenario in the absence of HIV/AIDS; and ‘Treatment’ is the scenario with HIV/AIDS and ART. Results are from the computable general equilibrium microsimulation model. GDP, Gross domestic product.

Despite the detailed costing of each expenditure item, a number of uncertainties remain. For example, unit cost estimates for ART using current Botswana drug prices and service models may prove too high in the longer term, although they are broadly consistent with current estimates [7]. There is also uncertainty about the future need and actual utilization of inpatient hospital care, and whether capacity will be developed to meet demand. Although we acknowledge these unavoidable limitations, we use the financial estimates in the CGE model to project government AIDS-related expenditures until 2021. In this way the economic models take into account not only the detailed structure of Botswana's economy, but also the demographic and financial projections.


Impact on economic growth

The models indicate that HIV/AIDS will have a substantial negative impact on the growth rate of the Botswana economy during 2003–2021 (Fig. 2). Comparing the With-AIDS and Without-AIDS scenarios suggests that Botswana's real GDP growth rate is declining by between 1.5 and 2% per year. This means that the economy will continue to grow despite high HIV prevalence rates, but it will be a third smaller in 2021 than it would have been in the absence of HIV/AIDS. This result takes into account projected reductions in labour supply, productivity, and investment. The scale of this impact on GDP is quantitatively similar to what would result if Botswana's mining industry was gradually shut down over the next 15 years, a sector that is currently the cornerstone of the economy. Although GDP declines as a result of HIV/AIDS, so too does population growth. Therefore, the estimated growth rate of GDP per capita falls by 0.5 to 1% a year, which is lower than the drop in aggregate GDP. Per capita GDP does decline, however, implying that the negative impact on GDP outweighs the decline in population (from 2.2 to 1.1% per year). As a result, average real incomes are 10–15% smaller in 2021 as a result of HIV/AIDS, implying that the remaining individuals and households are worse off because of HIV/AIDS (i.e. there is no ‘gift of the dying’).

Simulation results indicate that the government's treatment programme will mitigate some of the negative economic effects of the epidemic. ART raises GDP growth, adding a projected 0.4–0.8 percentage points to average annual growth over the next two decades (compared with the With-AIDS scenario). This results from a larger and healthier labour force, and a smaller effect on worker productivity. Providing treatment does not, however, eliminate all negative effects. Even under the Treatment scenario, HIV/AIDS still reduces economic growth by 1.2% a year compared with what would have been achieved had there been no epidemic. This means that after 20 years the economy is one-fifth smaller than it would have been. ART therefore offsets approximately a quarter to a third of the negative impact of HIV/AIDS on economic growth. Similarly, average per capita incomes still decline, but households are better off when ART is provided. These results are robust to different simulation methods, with both the macroeconomic and CGE models producing similar estimates of the reduction in growth caused by HIV/AIDS. The results are also within the range of previous studies for Botswana [14,15].

HIV/AIDS has a significant impact on Botswana's labour markets, worsening the employment situation. Although having a smaller population reduces the labour supply, lower investment and lower productivity reduce the demand for labour still further, leading to higher unemployment and slower wage growth than in the Without-AIDS scenario. Again, access to ART improves outcomes. At the sectoral level, the more labour-intensive sectors are hardest hit by the decline in labour force growth. These sectors include agriculture, manufacturing and the wholesale/retail trade. Agriculture is severely affected as it relies heavily on lower skilled labour, for whom HIV prevalence rates are higher and access to ART is lower. By contrast, the mining sector is less affected because growth in diamond production is forecast to slow regardless of the epidemic, and because the sector is highly capital intensive. Although there are uncertainties over productivity and investment impacts, varying the associated parameter assumptions does not lead to significantly different model outcomes.

Impact on the budget

HIV/AIDS will have a substantial effect on the fiscal budget, especially under the government's treatment programme that universally provides free ART. The total cost in 2006 is estimated at Pula 1.2 billion (US$200 million), equivalent to 6% of current government spending. These costs include: healthcare for inpatients and ambulatory patients; the ART programme; home-based care; prevention activities; care of orphans and vulnerable children; and additional old age pensions resulting from ART improving life expectancy (Fig. 3). Total costs under the treatment programme will increase in real terms by 60% by 2021, with ART accounting for approximately 40%. Incorporating these health costs into the CGE model, however, allows us to gauge how the faster economic growth resulting from treatment partly offsets the cost to the government. Although the share of AIDS-related health spending in total government spending will increase to 8% by 2011, it will begin to fall back to 7% by 2021.

Fig. 3
Fig. 3:
Projected government costs with and without treatment. (a) Projected AIDS-related costs to government under the treatment programme (i.e. Treatment scenario). (b) Projected AIDS-related to the government costs without treatment (i.e. With-AIDS scenario).

In the With-AIDS scenario, which excludes treatment, government health expenditures would be lower because the high cost of ART is avoided. These ‘savings’ would, however, be largely offset by higher costs to care for the sick and dying, including inpatient and ambulatory costs and home-based and orphan care. (This of course assumes that government policies and health system capacity could accommodate higher spending in these areas.) Furthermore, whereas total government spending on AIDS-related costs would be 15–25% lower, GDP and government revenues would also be lower because workers and their households, subject to higher morbidity in this scenario, would not be receiving treatment. Taken together, the financial projections and model results suggest that government health spending would be only 2% lower as a share of total expenditure if the treatment programme were not implemented. This finding is important, because it suggests that the seemingly prohibitive cost of providing treatment is deceptive once the cost of not providing treatment is calculated.

In order to maintain fiscal balance, higher AIDS-related spending must be matched by either reduced spending in other areas or else increased revenues. Reducing non-health expenditures would require successful prioritizing and budgetary controls. Alternatively, if AIDS-related spending is financed through larger budget deficits, then there will be upward pressure on interest rates and a crowding-out of private sector activity (i.e. slower economic growth). The model results indicate that funding AIDS-related spending through budget deficits is neither feasible nor sustainable, suggesting that tough spending trade-offs are unavoidable and will become more severe once ART is provided. It is important to re-emphasize that most AIDS-related spending will be required regardless of whether ART is provided. Furthermore, part of the additional ART costs may be met by donors, further reducing the impact on the overall government budget. In addition, this incremental spending on ART will help the economy to grow faster. The model results suggest that the additional spending on ART rising to Pula 360 million per year at present could lead to an additional GDP of as much as Pula 7.5 billion per year by 2021.

Impact on poverty

Poverty has been declining in Botswana over the past decade, albeit only slightly, and the model results presented in Figure 4 suggest that, even in the presence of HIV/AIDS, this trend will continue. The CGE model also indicates that the national poverty headcount would be only 1.5 percentage points lower in 2021 if there were no epidemic. Although this number may seem small, it suggests that even though the population would be substantially larger in the absence of HIV/AIDS, there would still be a smaller share of the population falling below the poverty line. The negative impact of HIV/AIDS results from slower income and employment growth, and the diversion of some household income towards health-related spending. The modelling also shows that providing ART can prevent between one-third and one-half of the poverty caused by HIV/AIDS. These results contradict assertions that HIV/AIDS reduces poverty because it disproportionately affects the poor, or that providing treatment worsens poverty by increasing the size of the population. Rather, the results indicate that the negative macroeconomic impact of HIV/AIDS in Botswana is sufficiently large to pull uninfected people and their households into poverty.

Fig. 4
Fig. 4:
Simulation results for poverty headcounts under the three scenarios. The poverty headcount is the share of population whose consumption expenditure falls below the dollar-a-day poverty line. Results are from the computable general equilibrium microsimulation model.


This paper has assessed the impact of HIV/AIDS on growth and poverty in Botswana, while also accounting for the potential effects of providing treatment. In the process we have incorporated demographic and financial projections within an integrated macro–microeconomic modelling framework. The results indicate that, over the next two decades, HIV/AIDS will have a substantial negative effect on GDP growth, per capita GDP, the poverty headcount, and government spending. Two key findings emerge. First, the macroeconomic impact of HIV/AIDS in Botswana is now severe enough to be affecting the economy as a whole, thereby pulling some of the uninfected population into poverty. Second, providing treatment causes only a marginal increase in health's share of total expenditure because of its positive effect on economic growth, and because other health services that have to be provided to infected people if they do not receive ART. These two findings confirm the importance of taking macroeconomic impacts into account in countries where HIV prevalence is particularly high.

Many of the policy issues arising from our analysis are simply an intensification of the broader economic challenges facing Botswana. In particular, there is a need to remove the obstacles to economic growth and diversification, to invest in training and education, and to stimulate employment creation. Reducing poverty and providing an effective social safety net are other imperatives. For these challenges, the macroeconomic and social policy framework that is already in place appears appropriate, although the epidemic makes overcoming challenges more difficult and the trade-offs between priorities more acute.

Some policy issues, however, relate specifically to HIV/AIDS and ART provision. Fiscal discipline and careful prioritization between health and other expenditures are still necessary in order to ensure fiscal sustainability. Particular care must be taken to maintain key investment spending, because the effects of HIV/AIDS on investment will most effect economic growth. Here, the government should fund AIDS-related spending by reducing public consumption rather than investment, and renew efforts to improve the climate for private and foreign investment. Reducing the cost of providing treatment can be as effective as increasing revenues. Here the government should work closely with donors to obtain access to concessional resources, while also exploring ways to reduce the costs of ART and increase the efficiency of its provision. The latter should include a review of the frequency of laboratory tests and consultations, as well as the mix of required staff, especially for patients who are stable on treatment. Also critical are referral systems and other means of optimizing the package of ART, home-based, inpatient and other care. Policy decisions in these areas, such as starting patients on ART sooner and increasing support for home-based care, may substantially influence the types of care that are needed and their associated costs. In particular, the government should explore ways to shift the inpatient burden from referral to district and primary hospitals, where beds are more available. In the labour market, the impact of HIV/AIDS in intensifying skilled labour shortages could be eased by encouraging labour immigration.

We have also compared two kinds of macroeconomic models in terms of their abilities to capture the impacts of HIV/AIDS. Our findings suggest that the growth and CGE models estimate broadly similar macroeconomic impacts. These outcomes depend first on population and labour supply effects, which can be measured with reasonable accuracy using demographic models, and second on estimated impacts on worker productivity, on which little research has been conducted to date. Even without large declines in productivity, however, the two models indicate that per capita incomes will fall in Botswana, which contrasts with a number of other studies. Moreover, the findings of the CGE model underline the importance of modelling the impact of HIV/AIDS at the sectoral and household level and with greater resolution than previous studies have employed. Whether HIV/AIDS disproportionately affects higher-skilled urban workers or lower-skilled rural workers influences a number of macroeconomic variables, such as aggregate savings and tax rates, which in turn determine the growth and fiscal impacts of HIV/AIDS. In the case of Botswana, this distributional effect proves to be important and pulls uninfected sections of the population below the poverty line.

Although we believe the conclusions of this study are robust, caution should be exercised when extrapolating estimated impacts to other countries because of the unique structure of Botswana's economy. Some of the issues raised, however, are likely to be relevant elsewhere. For example, in many countries with a high prevalence, the scale of the epidemic makes it important for Ministries of Finance, Health and Welfare to develop a better understanding of costs, economic impacts and fiscal sustainability. A comprehensive approach to understanding the economic impacts of HIV/AIDS is therefore necessary to inform policy and planning across ministries. Such an approach should not only look at service level costs, but also at a broader range of costs and benefits at sectoral and macroeconomic levels. Inaccurate or poorly understood costings of national HIV/AIDS strategic plans and frameworks can have serious negative consequences. Particularly in countries in which responses to HIV/AIDS are likely to be very costly, inadequate cost estimates may lead to wasteful budgeting, or priority services may not get sufficient funding. These effects will be most important in low-income countries with high prevalence rates.


The authors are grateful for comments from participants at the Conference on HIV/AIDS Intervention in Developing Countries: Use of Cost Effectiveness and Cost Benefit Analysis to Guide Policy and Action, held by the Harvard School of Public Health, September 2006.

Contribution of authors

K.J. conceptualized and managed the overall study and developed the aggregate growth model. A.K. developed the financial projections. H.S. analysed the AIDS and Household Income and Expenditure Surveys. J.T. developed the CGE and microsimulation model.

Sponsorship: Funding for the study was provided by the United Nations Development Programme in Botswana.

Conflicts of interest: None.


1. Bonnel R. HIV/AIDS and economic growth: a global perspective. South Afr J Econ 2000; 68:820–855.
2. Haacker M, editor. HIV/AIDS: the impact on the social fabric and the economy. In: The macroeconomic of HIV/AIDS. Washington DC: IMF; 2004.
3. Jefferis K, Kinghorn A, Siphambe H, Thurlow J. The economic impact of HIV/AIDS in Botswana. Report prepared for National AIDS Coordinating Agency and United Nations Development Program. Gaborone, 2007. Available at: Accessed: March 2008.
4. Cuddington J. Modelling the macroeconomic effects of AIDS, with an application to Tanzania. World Bank Econ Rev 1993; 7:173–189.
5. Cuddington J, Hancock J. Assessing the impact of AIDS on the growth path of the Malawian economy. J Dev Econ 1994; 43:363–368.
6. BIDPA. The macroeconomic impacts of the HIV/AIDS epidemic in Botswana. Report prepared for Government of Botswana. Gaborone: BIDPA; 2000.
7. Masha I. An economic assessment of Botswana's national strategic framework for HIV/AIDS. In: Haacker M, editor. The macroeconomic of HIV/AIDS. Washington DC: IMF; 2004.
8. Arndt C, Lewis JD. The macro implications of HIV/AIDS in South Africa: a preliminary assessment. South Afr J Econ 2000; 68:856–887.
9. Thurlow J. Is HIV/AIDS undermining Botswana's ‘success story’? IFPRI discussion paper 00697. Washington DC: IFPRI; May 2007.
10. CSO. Household income and expenditure survey 2002/03. Gaborone: Central Statistics Office; 2004.
11. Dorrington R, Moultrie T, Daniel T. The demographic of HIV/AIDS in Botswana. Report prepared by the Center for Actuarial Research for National AIDS Coordinating Agency and United Nations Development Program. Gaborone, 2006. Available at: Accessed: March 2008.
12. CSO/NACA. Botswana AIDS impact survey II. Gaborone: Central Statistics Office and National AIDS Coordinating Agency; 2005.
13. Fox M, Rosen S, MacLoed W, Wasunna M, Bii M, Foglia G, Simon J. The impact of HIV/AIDS on labour productivity in Kenya. Trop Med Int Health 2004; 9:318–324.
14. Botswana Institute for Development Policy Analysis. The macroeconomic impacts of the HIV/AIDS epidemic in Botswana. Report prepared for Government of Botswana. Gaborone: Botswana Institute for Development Policy Analysis; 2000.
15. MacFarlan M, Sgherr S. The macroeconomic impact of HIV and AIDS in Botswana. IMF Working Paper WP/01/80. Washington DC: IMF; 2001.

AIDS; antiretroviral therapy; Botswana; macroeconomic impact; poverty

© 2008 Lippincott Williams & Wilkins, Inc.