70584 v2 Georgia Programmatic Poverty Technical Note #2: Out of pocket payments for health JUNE 28, 2010 EUROPE AND CENTRAL ASIA REGION Out-of-Pocket Payments for Health in Georgia1 Summary: This note is part of the World Bank’s programmatic poverty work in Georgia, which is composed of a series of periodic notes on a range of policy topics of relevance to Georgia’s poor. It provides an overview of out-of-pocket payments (OOP) for health in Georgia, including the main causes and consequences of high OOP, with a special focus on pharmaceutical prices and the prospects for current policy initiatives in the area of health insurance to significantly lower OOP in Georgia. Three key policy messages that emerge are as follows: (i) much higher mark-ups for pharmaceutical products (a major source of OOP) are observed in Georgia compared to EU countries, and should be a focus of greater policy discussions; and (ii) the Medical Insurance Program for the poor has made a substantial contribution to lowering OOP among its beneficiaries, but there is scope to improve targeting and awareness in order to reach the many poor households not currently covered; and (iii) prospects for addressing OOP among the two-thirds of the population without adequate insurance coverage remain unclear, and may require additional public spending once the fiscal environment permits. 1. Introduction Out-of-pocket spending for health looms large in the lives of Georgia’s population, including the poor. When asked to identify the main problems faced by their family, about two-thirds mentioned the purchase of medicines, and nearly 60% cited (access to) medical services. In both cases these response rates were higher among the poorest quintile. The problems also appear to have grown worse between 2007 and 2009 (Figure 1). Along with food security and employment, these health-related concerns are at the top of the list of problems identified by Georgia’s population. More generally, when the Life in Transition Survey asked Georgian households to name the top priorities for government investment, the health sector was the most common answer, including among the poor. How to address out-of-pocket spending on health should thus be an important focus of policy discussions about improving the lives of the poor in Georgia. Health expenditures are different from most other household purchases. At least three reasons stand out. First, the spending is often not ‘voluntary’ (for example, if arising due to an unwanted health shock), and may have more to do with overcoming illness than adding to a household’s overall consumption or living standards, as is usually the case with other goods. Second, the uncertainty and potentially high cost associated with health expenditures make them amenable to prepayment and risk- pooling arrangements. Third, access to health care is often viewed as a ‘merit good’ which should not be determined primarily by ability to pay. For these reasons, a more desirable counterfactual to high out- of-pocket spending on health would be some form of pooling mechanism (whether through general taxes or a contributory insurance scheme), as well as cross-subsidization, to provide financial protection against health shocks. This has been achieved in many countries, but not yet in Georgia. 1 Prepared by Owen Smith, ECSH1. 2 Figure 1: Main problems cited by households, 2007-2009 70% 66.1% 60% 57.3% 57.7% 54.6% 52.0% 50% 44.9% 42.8% % of households 40% 33.3% 2007 30% 2009 20% 10% 0% Buying medicines Medical services Unemployment Hunger, undernourishment Source: WB LSMS and UNICEF WMS This policy note offers a closer look at the issue of out-of-pocket spending in Georgia. The next section gives an overview of trends, levels, decomposition and measurement issues. Section 3 speculates on the main causes of high OOP in Georgia. Section 4 discusses the consequences, including with regard to access to services and financial protection. Section 5 focuses on pharmaceutical products, the largest component of OOP. Section 6 analyzes current policy initiatives and their potential to address OOP. Section 7 concludes. 2. Overview of out-of-pocket spending in Georgia Out-of-pocket spending takes place in the absence of coverage of health expenses by state or private insurance schemes. Among a total population of about 4.4 million, there are about 950,000 beneficiaries of the Medical Insurance Program for the poor and an additional 200,000 public employees (teachers, police, etc.) with quite comprehensive coverage without co-payments, and who therefore face OOP only for non-covered expenses such as pharmaceuticals (although these are often significant). About 122,000 individuals have joined the new voluntary, publicly-subsidized 5-lari insurance program, but this has a narrower benefit package, meaning that significant OOP would remain for non-covered expenses, co-payments, and costs that exceed the program’s annual limits. About 250,000 individuals have private (typically corporate) insurance. Over two-thirds of Georgia’s population therefore only has access to the partially funded state programs. In principle, a significant share of out-of-pocket payments for the state programs is “informal�. That is, they do not correspond to a formal policy specifying levels of patient-side cost-sharing. However, to a significant extent these payments represent cost recovery on the part of providers, as opposed to pure 3 rent-seeking (i.e., extracting payments in excess of actual costs), chiefly because the budgets received by facilities (especially hospitals) do not cover the full cost of services covered by the state programs.2 More precisely, monthly allocations to providers from the state budget are typically used up well before the end of the month, after which providers would need to request payment from the patient for services nominally covered by the state program. Moreover, measuring the share of OOP that is “informal� is complicated by the fact that historically these programs have been poorly defined and are therefore not widely understood by the population or, indeed, by providers. Out-of-pocket spending has been the dominant source of health finance in Georgia over the past two decades. Table 1 shows the historical trend. While there has been some improvement (from a very high starting point), the share of OOP in total health expenditures in Georgia is still nearly twice the ECA regional average, and four times the EU-15 average. Table 1: Out-of-pocket spending as a share of total health expenditure, Georgia and region (%) 1995 2000 2005 2008 Georgia 92.5 77.8 76.9 66.3 ECA average 31.7 37.3 36.8 35.6 EU-15 average 17.7 17.2 16.6 15.7 Source: WHO The level of OOP in Georgia is subject to considerable measurement error (Box 1), but accounts for a major share of household expenditures. In 2009, the share of health expenditures in total household consumption was 9 to 11 percent across all quintiles (UNICEF). The most reliable estimate of annual per capita out-of-pocket spending on health was GEL 216.2 as of 2007, or about 5 to 6 percent of per capita GDP. A decomposition of OOP into subcategories highlights the importance of drug spending in OOP (Table 2). Estimates of relative shares are quite consistent across surveys, with medicines accounting for roughly half of all OOP, and hospital services and ambulatory services both representing about 20-25 percent of the total. Although drugs represent the largest category of spending, the variability of OOP on health services is larger. In addition, these shares vary by socioeconomic status, with poorer households spending over two-thirds of their OOP on drugs, while the corresponding figure for the better-off is closer to 40 percent. The share of spending devoted to hospital and surgical services is quite low by the poorest quintile, but much higher by richer quintiles. Table 2: Out-of-pocket spending by category, per person, per year (2007) Type GEL per capita per year In-patient care 39.3 Out-patient care 41.2 Drugs 104.6 Other 31.1 Total 216.2 Total as share of per capita GDP 5.6% Source: HUES 2 Euro Observatory, Georgia HiTS, 2009. A full analysis would require a detailed costing study, as recently done in Armenia. 4 The characteristics of households that face high out-of-pocket health spending provide an indication of who is most vulnerable. A regression of the probability of suffering catastrophic OOP (defined as exceeding 10 percent of total consumption) on various household characteristics suggests that the most vulnerable are those households with elderly (over 65), young children (under 5), and a household head who is not employed. The education and gender of the household head and urban/rural location are not significant correlates of high OOP. Box 1: Measuring OOP: How high is out-of-pocket (OOP) spending on health in Georgia? Estimating out-of-pocket payments through household surveys can be challenging. A common approach is to ask individuals about their last visit to a health care provider and the associated expenditures, often restricted to a certain time frame such as the past month. However, those who are sick, especially with a chronic illness, may make several visits during that period, and often to more than one provider, with expenditures at each stage. The results in this case will tend to underestimate expenditures. Also, rare events such as hospitalizations are more easily captured through longer recall periods, whereas more regular occurrences such as drug purchases or clinic visits are better addressed over shorter periods to reduce recall bias. In addition, surveys may ask about drug purchases in association with a provider visit, whereas spending may occur in isolation from a health care episode (e.g., self-treatment). For these reasons, guidelines for health modules tend to recommend a more thorough survey to capture all visits, and multiple recall periods depending on the type of service (for example, one month for outpatient care and 12 months for inpatient care)3. But an extensive health module may not be feasible in the context of a general consumption expenditure survey. The regular household budget survey (HBS) in Georgia may be subject to some of these shortcomings in estimation. The HBS uses a four-week recall period for utilization that can capture only one visit for general curative care, and does not ask about utilization at all for the chronically ill. Expenditures are captured using a three-month recall period. The UNICEF Welfare Monitoring Survey (2009) used a one- year recall for health expenditures. On the other hand, the 2007 Georgia Health Utilization and Expenditure Survey (HUES) was able to focus exclusively on health issues and thus employed a survey instrument that more closely followed the recommended approach. As a result, there is considerable variation across surveys in their estimates of OOP for health in Georgia. While the HBS survey estimated average health expenditure of about GEL 70 per person annually in 2007, the HUES estimate in the same year was three times higher. The UNICEF survey of 2009 was in between, at about GEL 120 per person annually. Although the true figure is unknown, it is likely that the HBS results are significantly underestimated. In 2010, a second round of the HUES and UNICEF surveys (in the latter case, with a revised health module) will take place. An effort is also under way to revise the HBS health module with World Bank technical input. 3 Gertler et al., 2000. 5 3. Causes of high OOP One of the main historical reasons for high out-of-pocket health spending in Georgia has been relatively low government expenditure on the sector. Figure 2 shows the international pattern. The corollary of this inverse relationship is that private pooling or voluntary health insurance rarely offers a solution to high OOP. Very few countries in the world have successfully pooled more than 10 percent of total health spending through voluntary private means, and in all cases this was achieved primarily for higher income groups.4 This issue is discussed further below. In effect, total (public and private) health expenditure as a share of GDP typically falls in a relatively narrow range between 6 and 8% (albeit higher in rich countries), and this is divided between public (whether general-tax or payroll-tax financed) and OOP sub-components largely on the basis of how much the government spends. In other words, after government spending is taken into account, most of the rest will fall to OOP. However, Figure 2 also indicates that at any given level of government health expenditure, there is some variation in reliance on OOP. This is likely to depend on other factors, such as utilization rates (in part reflecting epidemiology and demographics) and input prices. Georgia has one of the lowest utilization rates of health care in ECA, so this is not a key driver of high OOP. However, the high price of pharmaceuticals in Georgia is an important factor that is discussed in detail below. While other factors are also at play, the high price of inputs (such as drugs) is partly a result of low government spending, because the bargaining power associated with only having one (or few) purchasers of medical care is foregone. Figure 2: Government health spending and OOP expenditures around the world Government health spending and OOP expenditures around the world 100% 90% OOP as % of total health expenditure 80% Georgia 70% 60% 50% 40% 30% 20% 10% 0% 0 1 2 3 4 5 6 7 8 9 10 Government health spending as % of GDP 4 WHOSIS database. 6 In principle, low government spending on health can be a result of either a small government sector, or due to a relatively small allocation to health within the existing fiscal envelope. However, a small government does not appear to be the applicable story in Georgia, as total government outlays as a share of GDP were 35% in 2009, which is close to the regional ECA average. A more likely explanation is the allocation within the fiscal envelope, as only 5.3 percent of the 2009 budget was spent on health, the second lowest share in the region. Historically, a small budget allocation to priority sectors such as health has been in part due to relatively high spending on both defense and public order and security (based on IMF classifications) compared to the rest of ECA. However, since 2008 this has begun to change and the budget share allocated to health has started to rise (to 6.4 percent in 2010). In general, however, prospects for further increases in the health budget over the medium-term will be challenged by the need for fiscal consolidation in the difficult post-crisis environment. 4. Consequences of high OOP A health financing system that relies heavily on out-of-pocket payments, such as Georgia’s, poses challenges for the achievement of both of the two major goals of a health system. These are to improve the level and distribution (equity) of health outcomes, and to provide financial protection against high costs associated with health care. Facing financial barriers to access health care is a common occurrence in Georgia. Figure 3 shows the share of respondents reporting that they did not seek care or fill a prescription “because it was too expensive�. The problem is more common among the poor. Since these events are self-reported, it is not certain that all such visits or hospitalizations were medically necessary. However, utilization rates of both in-patient and out-patient services in Georgia are about one-quarter to one-third of the ECA averages.5 Thus, under-utilization rather than over-utilization should be of greater concern. Figure 3: Financial barriers to care, by quintile (Did not seek care “because it was too expensive�) 30 25 20 % of respondents 1 15 2 3 4 10 5 5 0 Did not seek care when sick Did not fill prescription Was not hospitalized when in last 30 days following consultation needed in last year Source: HUES 2007 5 WHO Health for All database. 7 High out-of-pocket payments for health may also cause a household to fall below the poverty line – that is, they can be “impoverishing�. If a household has total consumption expenditures (pre-OOP) above the national poverty line, but their total non-medical spending (post-OOP) is below the poverty line, they could be considered to have suffered impoverishment due to OOP for health. Figure 4 shows this graphically based on 2007 data. Households are ranked along the horizontal axis by total consumption. The vertical drip lines represent OOP for health, and the poverty threshold is indicated by the horizontal line. Applying this approach to 2007 household survey data, it has been estimated that an additional 3 percent of Georgian households were poor as a result of OOP for health.6 Consistent with Georgia’s high reliance on OOP as shown above, this level of poverty impact is notably higher than most other countries either in the region or around the world. 7 Figure 4: Impoverishment due to out-of-pocket payments, 2007 An alternative approach for highlighting the impact of OOP on households is to measure the extent to which they are “catastrophic.� Impoverishing OOP puts the emphasis on crossing the poverty line irrespective of the size of payments. Catastrophic health expenditures occur when they exceed some threshold of either total or non-food expenditure. The choice of threshold is somewhat arbitrary, but a common practice in recent literature uses 10 percent of total consumption expenditure or 25 percent of nonfood expenditure. The share of households with OOP exceeding 10 percent of total expenditure in 2007 was estimated to be 17.6 percent in Georgia, while the share of households for which OOP exceeded 25 percent of non-food expenditure was 25.7 percent. A third definition of catastrophic OOP, exceeding 40 percent of “capacity to pay�, was used to calculate results for 59 other countries, and Georgia has a higher incidence than all but 6 of these, and all but 1 of 13 transition economies (Table 3). 6 Whether this is an accurate way to evaluate the true poverty impact of OOP is a matter for debate, since in reality households are likely to draw on several possible coping mechanisms that would allow for consumption smoothing, such as drawing down savings, borrowing, or selling assets. Thus, a costly illness episode in one period would not necessarily have an immediate and commensurate impact on total consumption in the same period. Nevertheless, the concept can be useful for international comparisons of financial protection. 7 World Bank (2009). Georgia Poverty Assessment. 8 Table 3: Share of households with OOP for health exceeding 40% of “capacity to pay� Country % of households Azerbaijan 5.8% Bulgaria 2.0% Croatia 0.2% Czech Republic 0.0% Estonia 0.3% Georgia 5.1% Hungary 0.2% Kyrgyz Republic 0.6% Latvia 2.7% Lithuania 1.3% Romania 0.1% Slovakia 0.0% Slovenia 0.1% Ukraine 3.9% Western Europe average 0.6% Source: LSMS calculations for Georgia; otherwise Xu et al., 2003. In sum, although the disruptive impact of OOP on Georgian households is difficult to quantify with precision, there is strong evidence pointing to a lack of financial protection in health in Georgia. OOP for health is high whether measured relative to total consumption or by international standards. Section 6 considers the scope for current policy initiatives to help in this regard. Box 2: Welfare consequences of high OOP: Basic principles and an application to US Medicare The trade-off between risk protection and moral hazard is one of the classic issues addressed in health economics. A risk-averse individual would prefer the certainty of paying a fixed insurance premium to facilitate consumption-smoothing rather than run the risk of facing a very high out-of-pocket payment in the event of a health shock. However, reducing the cost of health care to zero (or close to it) at the point of service may invite over-utilization of care (i.e., services for which the marginal benefit is lower than marginal cost). “Optimal health insurance� would balance these two factors. However, behavioral economics has posed a challenge to the traditional framework and suggests that greater subsidization of health care may be warranted.8 This line of reasoning argues that decision-making capacities of individuals are far from optimal, and that they have difficulty making choices in the face of uncertainty, balancing short-term costs with long-term benefits, and processing complexity – all of which are common in the arena of health care consumption. A more complete ‘balance sheet’ of the welfare implications of a health financing arrangement would weigh several factors. These include the benefits of financial risk protection, the cost of moral hazard, the marginal cost of public funds, and the value of the health improvements attributable to the health financing intervention. This is a difficult undertaking, but has been done for the introduction of Medicare in the US in 1965.9 In its first five years, Medicare reduced out-of-pocket spending by about 40 8 Liebman, J., and R. Zeckhauser (2008). “Simple Humans, Complex Insurance, Subtle Subsidies,� (NBER WP14330). 9 Finkelstein, A. and R. McKnight (2008). "What Did Medicare Do? The Initial Impact of Medicare on Mortality and Out of Pocket Medical Spending." Journal of Public Economics 92: 1644-1669. 9 to 50 percent among the top quartile of the OOP distribution. Using an expected utility framework, this was estimated to yield welfare gains via consumption-smoothing equivalent to about 40 percent of the social cost of Medicare (i.e., efficiency losses due to moral hazard and the cost of revenue-raising). Given the very high value attached to better health, even small mortality and morbidity improvements due to Medicare would translate into an overall favorable balance. While the modern version of Medicare has many pros and cons, the key message from this study is that the welfare gains arising from reducing OOP are substantial. 5. A closer look at pharmaceuticals As noted, expenditures on pharmaceuticals account for about half of total out-of-pocket spending in Georgia, and about two-thirds among the poorest quintile. The purchase of medicines is also one of the most commonly cited problems faced by households. The share of total health expenditures spent on drugs is therefore about 35-40 percent (the government also spends a small amount on drugs for specific programs), which is well over twice as high as the corresponding rate among most OECD countries (typically about 15 percent).10 This is partly due to high levels of self-treatment, as the population directly visits pharmacies in order to avoid paying out-of-pocket at health facilities. Also, it is reported that vitamins and food supplements are among the top selling pharmaceuticals, often with uncertain therapeutic benefits. But high prices are almost certainly a part of the picture as well. A major source of recent concern, and a focus of growing public debate, has arisen from a perception that drug prices in Georgia are high and rapidly increasing. Figure 5 shows the recent inflation trend for drugs, with a sharp spike in late 2008 followed by a gradual downward trend. Access to affordable medicines is an issue for the whole population, as most out-patient drugs are not covered by the universal benefit package, and were only added to the Medical Insurance Program for the poor in 2010 (and with a ceiling on reimbursement). High OOP spending for drugs is thus a key factor underlying Georgia’s weak performance on financial protection described in the previous section. Meanwhile, evidence from advanced countries suggests that pharmaceuticals (e.g., anti-hypertensives, cholesterol drugs) have played a major role in extending life expectancy. Thus, access to medicines is an issue with relevance to health outcomes, poverty, inflation, and governance. 10 OECD Health at a glance 2009. 10 Figure 5: Inflation of pharmaceuticals in Georgia, year-over-year, 2007-09 35 30 25 20 15 10 5 0 -5 Mar-07 Mar-08 Mar-09 May-07 May-08 May-09 Dec-07 Dec-08 Dec-09 Oct-07 Oct-08 Oct-09 Jan-07 Jul-07 Jan-08 Jul-08 Aug-07 Jan-09 Jul-09 Apr-07 Aug-08 Aug-09 Apr-08 Apr-09 Feb-07 Feb-08 Feb-09 Sep-07 Sep-08 Sep-09 Jun-07 Jun-09 Nov-07 Jun-08 Nov-08 Nov-09 Source: GeoStat Medicines Overall CPI The structure of Georgia’s pharmaceutical market and recent developments in the country appear to be major underlying reasons for the public concern. Two pharmaceutical companies control a significant portion of the pharmaceutical supply chain (production, import, wholesale and retail network) and consequently control up to 70 percent of a pharma market estimated to be worth about GEL 450 to 500 million in 2007 (about US$270m to 300m).11 A third company is also a significant player. The two main companies also have interests in the hospital and health insurance industries. Because there is little coverage of drugs by the various state programs, opportunities for using the leverage of a large single purchaser to lower prices are foregone. Instead, individuals with weak bargaining power must pay out-of-pocket themselves. Against this background, and because of the lack of hard evidence on drug prices in Georgia, a survey of retail pharmacy prices and availability for 50 common drugs was conducted in December 2009 with World Bank support.12 A total of 146 pharmacies were surveyed across six regions of Georgia (Tbilisi, Batumi, Kakheti, Samegrelo, Kvemo Kartli, and Imereti). It included pharmacies in the three major networks, independent pharmacies, and hospital pharmacies. Price data was collected for both brand name and their cheapest generic equivalents. An analysis suggests that across geographical regions and pharmacy types (network vs. independent), there is little variation in pharmaceutical prices in Georgia. To provide a benchmark, price comparison data for a list of 20 of those drugs was obtained for six European Union countries.13 Over 90 percent of pharmaceuticals in Georgia are imported. Figure 6 shows some indicative results for net pharmacy retail prices (most EU countries apply a VAT to drugs, 11 WHO (2009). Georgia Health in Transition Series. Citing MoLHSA National Health Accounts. 12 Fieldwork was undertaken with support from a Norwegian trust fund for assessing pharmaceutical governance. 13 EU data was obtained from Gesundheit Osterreich (OBIG) and its Pharmaceutical Pricing and Reimbursement Information service (PPRI), also with partial trust fund support. 11 but Georgia does not). Clearly salary and retail property costs would be more expensive in the (relatively rich) EU countries. The numbers suggest there is wide variation in price differences between Georgia and the EU countries (Greece, Italy, Czech Republic, Hungary, and Poland). A general pattern, however, is that original (or reference brand) drugs are typically more expensive in Georgia than in the EU, often by a wide margin, while most generic drugs are cheaper in Georgia. This is consistent with reports that both primary care doctors and pharmacists receive payments from the pharmaceutical companies in order to encourage the population to purchase specific (higher-cost) drugs. The population, meanwhile, may distrust lower-cost drugs. Figure 6: Difference in retail drug prices in Georgia vs. five-country EU average 150% 100% 50% % difference 0% Brand Generic -50% -100% Source: WB survey and OBIG In addition to price comparisons, an analysis of mark-ups along the supply chain also provides important insights into market dynamics. Figure 7 shows estimates of the combined wholesale and retail margins for Georgia and five EU countries. The Georgian data is based on analysis of customs data, while EU data reflects actual ex-factory, wholesale, and retail prices obtained from OBIG/PPRI. The Georgian average is clearly much higher than in other countries. Because drugs are high-value, low-volume products, these large differences are unlikely to be caused by inefficiencies in the distribution system. The Georgian retail margin is reportedly quite small, as they are reduced to a minimum by wholesalers (where the market is most concentrated). A major factor behind these results is that pharmacy margins (in some cases both wholesale and retail) are regulated in all 27 EU member states. Typically they take the form of either a linear mark-up or a regressive scheme (higher mark-ups for cheaper drugs, to encourage pharmacists to dispense lower-cost formulas). The results in Figure 7 for the EU countries reflect such policies. However, these are indicative results for fewer than 50 drugs, and thus represent a sub-set of the market requiring further exploration. 12 Figure 7: Estimated average combined wholesale and retail margins, Georgia and five EU countries 120% 102% 100% 80% % margin 60% 49% 46% 40% 35% 32% 27% 20% 0% Georgia Czech Rep. Greece Hungary Italy Poland Country Source: WB survey and OBIG The government has begun to take pro-active measures to address the issue of pharmaceutical prices. In November 2009, a law was passed which aimed to reform certain aspects of pharmaceutical policy in Georgia. This included three main objectives: (i) to ease the procedures for importing drugs already included on the approved lists in advanced countries; (ii) to make it easier for new market actors to import drugs directly (e.g., hospitals, insurers, etc.); and (iii) to loosen the requirements for retail pharmacy space (e.g., enabling supermarkets and others to sell medicines). The impact of these measures on pharmaceutical prices remains to be seen but will be monitored in the months ahead. In April 2010, a new out-patient drug benefit was added to the Medical Insurance Program for the Poor. This provides an annual subsidy up to GEL 25 for coverage of pharmaceuticals per beneficiary, with a 50 percent patient co-payment (i.e., up to GEL 50 purchasing power). Only generic drugs are reimbursed in order to ensure lower prices, and a prescription is required to encourage use of primary care and rational drug use. While its flat, per capita design places a ceiling on “insurance� (i.e., transfers from healthy to sick), it is an important initiative and offers the potential to replicate the successful example of Kyrgyzstan’s out-patient drug benefit. Since reimbursement involves a role for insurers, providers, government, and patients, implementation arrangements should be closely monitored. 6. OOP and current major policy initiatives Two major health insurance initiatives in recent years have been aimed in part at addressing the problem of OOP in Georgia. These are the Medical Insurance Program for the poor, launched in 2006 and revised several times since, and the new, voluntary “affordable 5-lari� health insurance program started in 2009 and aimed at the rest of the population. 13 The Medical Insurance Program for the poor offers a comprehensive benefit package to eligible households. Most emergency out-patient care and planned or emergency in-patient care is included, with few coverage limits and no co-payments. Most of the non-MIP population only has access to the universal basic package. This offers a more limited package of services with significant co-payments, but suffers from a lack of clear definitions, chronic under-funding, and widespread informal payments. Eligibility for the MIP program is determined by the same proxy means test used for the targeted social assistance (TSA) program, but with a higher eligibility threshold. The proxy means test includes over 100 indicators and is administered by the Social Services Agency (SSA) to any household that applies (about 40 percent of the population has applied). The state budget covers all households up to a score of 70,000, while two regions, Adjara and Tbilisi, also covered households with scores between 70,000 and 100,000 during 2008 (Adjara no longer does). In 2009 the MIP covered 900,000 beneficiaries – about 20 percent of the population – and had a budget of GEL 127m, or roughly half the state health budget. To assess the impact of MIP on key outcomes of interest, an impact evaluation based on a regression discontinuity design was undertaken in 2008-09. The sample was about evenly divided into MIP and non-MIP beneficiaries. In order to ensure that the MIP participants who were surveyed – the “treatment� group – could be compared with the most similar possible “control� group of non-MIP households, the survey only collected data on households just above and just below the eligibility thresholds (70,000 or 100,000). Box 3 has further details. Box 3: Regression discontinuity design for evaluation of MIP The MIP evaluation survey covered 3,600 households (about 11,000 individuals) across most regions of Georgia, and asked about their health utilization and expenditures, their health status, and knowledge and satisfaction of MIP beneficiaries with the program. The sample was about evenly divided into MIP and non-MIP beneficiaries and only collected data on households just above and just below the eligibility thresholds (70,000 or 100,000 depending on the region). Trying to identify MIP’s impact by comparing households with scores of 40,000 and 120,000, for example, would be very difficult because they are different in many other ways besides their MIP status. This approach, called “regression discontinuity design�, is an increasingly common technique used to conduct impact evaluations around the world. In essence, it is based on the assumption that a household with a score of, for example, 69,000 is almost identical to a household with a score of 71,000, except for the fact that one received MIP and the other did not. Once the data was collected, it was confirmed that the MIP and non-MIP households in the sample were nearly identical in all respects (e.g., demographic characteristics, number of household assets, etc.) that might affect the outcomes we are interested in, except for whether or not they were enrolled in MIP. This gives us confidence that the data can be used to make an accurate assessment of MIP. Figure 8 shows the significant impact of MIP on out-of-pocket expenditures for health care. For outpatient care in Adjara and Tbilisi, and inpatient care in all regions, MIP beneficiaries pay approximately 50 percent less than non-beneficiaries (there is no statistically significant difference for outpatient care in the regions with a cut-off score of 70,000). The survey also found that MIP beneficiaries were more likely to report receiving free or reduced-price care because of insurance, and less likely to report that they could not pay for the costs of care out of their usual income. Together these findings indicate that MIP has made a major contribution to reducing out-of-pocket spending among its beneficiaries, and is therefore achieving one of its key program goals. The survey results also indicated, however, that the program has not had any impact on utilization. This is an important finding, but here we focus on the issue of financial protection and OOP. 14 Figure 8: Out-of-pocket spending per episode, GEL 700 596 600 500 431 400 GEL 300 221 233 200 100 85.2 63.6 52.2 43.6 0 Out-patient 70,000 In-patient 70,000 Out-patient 100,000 In-patient 100,000 MIP Non-MIP Source: Bauhoff, Hotchkiss, and Smith, 2010 The main reason why out-of-pocket spending has not fallen to zero among MIP members is because of drug expenditures. The study was undertaken prior to the recent introduction of an out-patient drug benefit as part of MIP. However, some respondents also reported paying for certain services that are supposed to be covered by MIP, indicating that informal payments may persist and there is scope to improve knowledge of the benefit package. While the initial impact of MIP on OOP has been substantial, some caveats are in order about its potential to bring Georgia significantly closer to international averages for OOP reliance. First, a very rough calculation would suggest that a 50 percent reduction of OOP among the 20 percent of the population who are MIP beneficiaries has reduced OOP by no more than 10 percent overall (and probably less since MIP beneficiaries are poorer than average). Yet as already noted, OOP reliance in Georgia is about twice the regional ECA average. Thus, it raises the question of how to improve risk protection for the other 80 percent of the population (discussed below). Invariably, increased government spending within a sustainable fiscal framework will be required, although the need for fiscal consolidation over the medium-term will make this difficult to achieve. In addition, although MIP represents a best-practice example of targeting limited public funds to the poor, the actual targeting performance could be improved (Figure 9). This is in part because the proxy means test used to determine MIP eligibility was originally developed for the TSA program, meaning that it was intended to identify the extreme poor (it is budgeted to cover about 450,000 beneficiaries), rather than to identify a much larger group of beneficiaries as in the case of MIP (which is budgeted to cover over 900,000 individuals). As a result, due to sub-optimal targeting performance, nearly 70 percent of MIP beneficiaries are not in the bottom quintile, and only 28 percent of those who are in the bottom quintile are covered. Another factor is that only 35-40 percent of the Georgian population has applied for inclusion in the database of socially vulnerable families. A better understanding of why many poor households have not applied may also lead to better coverage of the lowest quintile. 15 Figure 9: MIP targeting % of MIP benefits received by each quintile 35% 30% 25% 20% % 15% 10% 5% 0% 1 2 3 4 5 Source: UNICEF WMS Quintile For the population not covered by MIP, the government introduced a voluntary, publicly co-financed “5-lari� health insurance program in March 2009. This offers a subsidized minimum benefit package covering primary care, some diagnostic tests, acute hospitalizations (e.g., appendicitis), and emergency care. Certain co-payments and annual limits apply. Planned in-patient care and drugs are not covered. The premium is GEL 5 per beneficiary per month, with two-thirds funded by the government and one- third by the individual. At present, coverage is 122,000 individuals (about 3 percent of the population) who signed up during the open enrollment period between March and July 2009. A new enrollment period is tentatively planned for mid-2010. At this early stage, few conclusions can be drawn about the impact of the 5-lari program. Preliminary evidence from claims data indicates that utilization rates are low, but this needs to be corroborated with survey data.14 Higher uptake and service use would be required in order to have an impact on overall OOP. As noted, there are virtually no examples of low and middle-income countries that have succeeded in significantly lowering OOP among lower income quintiles through voluntary health insurance. The government’s subsidization of Georgia’s 5-lari program offers the potential of encouraging uptake (a lack of trust of private financial institutions has been hypothesized as a reason the poor do not buy insurance), but this remains to be proven. Other policy initiatives in the sphere of health insurance have been floated, but these have yet to be launched. One option would be to combine those state programs (e.g., cardiology, oncology) which are insurable and create a publicly-subsidized insurance package for the elderly. Key issues to address include the comprehensiveness of the benefit package and the level of public funding that can be committed in the current fiscal environment. 14 Sirois, A. “An Evaluation of the new voluntary health insurance program in the Republic of Georgia�. Draft. 16 7. Summary and policy issues Key facts related to out of pocket payments for health can be summarized as follows.  Paying for medicines and accessing medical care are among the most common problems reported by Georgian households.  Out-of-pocket spending on health accounts for about 10 percent of total household consumption, and Georgia’s reliance on OOP (measured as a share of total health expenditures) is about twice the ECA average and four times the EU-15 norm.  Historically, one of the main causes of high OOP in Georgia has been relatively low government spending on health, both as a share of GDP and as a share of the government budget.  The prevalence of OOP as the main mechanism for health finance contributes to one of the lowest utilization rates of health care in the region, and one of the highest rates of “impoverishing� and “catastrophic� health care payments by households. Key policy messages include the following:  Increased government spending on health would help reduce OOP, but this option is constrained in the current fiscal environment. Re-allocation from other sectors may be an alternative.  About half of total OOP is for pharmaceuticals, and drug prices in Georgia do not compare favorably with EU prices, in large part due to very high mark-ups. Current policy initiatives in the pharmaceutical sphere – the new legislation and the MIP out-patient drug benefit – should help. Other measures may also merit consideration. In all EU countries and in many others, wholesale and retail price margins are regulated. If this approach is adopted, it would need to be accompanied by strong enforcement efforts. Close monitoring and scrutiny of price developments by the competition commission may also be warranted. Encouraging buyer cooperatives among independent pharmacies to improve their bargaining position vis-à-vis wholesalers is another possibility. On the patient side, there is a need to educate and reassure the population about the quality of lower-cost drugs.  The Medical Insurance Program for the poor has had a major impact on OOP among beneficiaries. The program could be strengthened further by better tailoring the proxy means test that determines eligibility to the target population, and encouraging take-up among those who have not applied. Meanwhile, ongoing oversight of program implementation by private insurers and providers (including through the health insurance mediation service) will help support a continued positive program impact.  Lastly, international experience suggests the voluntary 5-lari insurance program will face significant challenges in enrolling a large share of the population, particularly the poor. Further monitoring and, if necessary, experimentation with program design will be important. Potential next steps for the World Bank’s Georgia programmatic poverty work that emerge from the foregoing discussion include the following:  Broader dissemination and discussion of findings from the pharmaceutical price survey.  Technical inputs to review and improve the proxy means test methodology as it applies to determining MIP eligibility, and identifying possible causes for non-application to the database for vulnerable families by certain segments of the population. 17