HEALTH ECONOMICS Health Econ. 24: 1560–1572 (2015) Published online 2 October 2014 in Wiley Online Library (wileyonlinelibrary.com). DOI: 10.1002/hec.3111
DOES PRESUMED CONSENT SAVE LIVES? EVIDENCE FROM EUROPE ZEYNEP BURCU UGUR* Department of Economics, Tilburg University, Tilburg, Netherlands
SUMMARY One policy tool that could affect organ donation rates is legislative defaults. In this study, we examine how presumed consent impacts cadaveric donations and kidney transplantations, using a panel dataset from the EU-27 countries plus Croatia in the period 2000–2010. We find that presumed consent countries have 28% to 32% higher cadaveric donation and 27% to 31% higher kidney transplant rates in comparison to informed consent countries, after accounting for potential confounding factors. After studying willingness to donate one’s organs and registering preferences for organ donation, we find that presumed consent could increase cadaveric donation rates, because people fail to register their preferences and many have no preference for organ donation. Copyright © 2014 John Wiley & Sons, Ltd. Received 22 January 2014; Revised 10 August 2014; Accepted 8 September 2014 JEL Classification: KEY WORDS:
I18; K32; D82
presumed consent legislation; organ donation; kidney transplantations
1. INTRODUCTION The question of how to generate more organ donors is highly relevant, because the chronic shortage1 of organs is directly linked to the death of many patients, both in Europe and elsewhere. One policy option is to adjust legislative defaults. There are two types of legislation for obtaining consent for organ transplantation: informed consent and presumed consent. In informed consent, individuals declare their willingness explicitly by registering to be an organ donor. In presumed consent, a brain-dead individual whose organs are suitable for transplantation is automatically considered to be a donor unless she has stated a preference for not donating.2 Theoretically, if the costs of registering preferences for organ donation are low,3 the defaults would not have a great effect for fully rational individuals who already have established preferences for organ donation. In the case of a mismatch between the default and the preferences, individuals are expected to take action for the desired option. However, if individuals are more likely to accept the effortless default option rather than make a choice that has mental costs, defaults might matter. The mental costs of making a decision for organ donation could be substantial, because it requires thinking about one’s own death, which is generally perceived as *Correspondence to: Department of Economics C214, Hukumet Meydani No: 2 Ulus, Social Sciences University of Ankara, Ankara, Turkey. E-mail: [email protected] 1
See Appendix A for more information on the extent of the organ shortage problem in Europe. In practice, in some presumed and informed consent countries, consent from the family of the deceased is routinely sought even if he or she explicitly stated their preference to be a donor. 3 There may be differences in registration costs between countries: some countries allow for registering organ donation preferences through online forms, others have dedicated telephone lines. Some only accept applications by mail. Some only accept registration at specific locations, which may require incurring transportation costs. 2
Copyright © 2014 John Wiley & Sons, Ltd.
PRESUMED CONSENT
1561
unpleasant and stressful. Thus, there might be differences in terms of the age structure of the donor pool between presumed and informed consent systems. Young people, whose organs are more valuable for transplantation, are less likely to think about death and register as organ donors, because they may anticipate a long life. Presumed consent could solve the registration problem for young people. Even if presumed consent is not strictly enforced, in that family consent is always sought, legislating for presumed consent might still be a positive signal from the government to the families. Presumed consent legislation might even impact the way in which doctors talk to families. Doctors then have the law on their side when trying to explain the need for organ donation to a bereaved family. Some studies find higher organ donation rates after the enactment of presumed consent legislation. After the introduction of presumed consent in Austria in 1982, the donor rate quadrupled (Gnant et al., 1991). Similarly, Roels et al. (1991) reported more than a doubling of kidney donations in Belgium after the introduction of presumed consent in 1986.4,5 Some studies (Johnson and Goldstein, 2003; Abadie and Gay, 2005; Bilgel, 2010) find higher organ donation rates in presumed consent countries compared with informed consent countries. However, the evidence is not entirely convincing. For instance, Abadie and Gay (2005) acknowledged that if presumed consent is enacted in countries where there is higher social acceptance of organ donation, the found effect might be biased. Bilgel (2010) used the fixed effects vector decomposition, which is not econometrically valid (Greene, 2011). Moreover, there is no consensus in the literature on whether the presumed consent legislation is an indicator of a country’s commitment to organ donation or a causal mechanism in itself (Healy, 2005). Yet, according to medical professional opinion,6 the most effective option for alleviating organ shortage is enactment of presumed consent (Oz et al., 2003). The other potential channels7 do prompt potential donors to take action but do not significantly influence donation decisions in the population as a whole. This study contributes to the literature in some important ways. We firstly address potential endogeneity of presumed consent by showing evidence that presumed consent is not necessarily legislated in countries where there is higher public support for organ donation. Secondly, we address the claim that presumed consent is an indicator of a country’s commitment to organ donation rather than a causal mechanism in itself. We show that after taking into account a country’s commitment to organ donation, proxied by kidney transplant centers as an additional control variable, the coefficient of presumed consent is still statistically significant and even increases. Thirdly, although according to the Eurobarometer 2009 survey, religious reasons,8 distrust in the system and fear of manipulation of the human body are three major causes of refusing organ donation, previous studies have not dealt adequately with trust in the system and with religious differences.9 To capture trust in the system, we included corruption perception scores from Transparency International. To control for religiosity changes over time, we compiled the percentage of people being Roman Catholic10 and having no religion, from various surveys. Lastly, to the author’s knowledge, this study is the first to analyze the impact of presumed consent on kidney transplantation, which is of particular relevance from a policy perspective.
4
Legislative change in Austria and Belgium is not used in this study, because these changes occurred before international country-level organ donation data were available. 5 The International Registry of Organ Donation and Transplantation provides organ donation rates from 1993 and transplantation rates from 1999. 6 The results are based on a survey of 739 members of the International Society for Heart and Lung Transplantation on improving organ donation levels. 7 See Appendix B for discussion on other policy options for promoting organ donation. 8 Although the Roman Catholic Church explicitly supports organ donation, some denominations such as Jehovah’s Witnesses and the Christian Scientists discourage these practices. In Islam and Judaism, organ donations and transplants are allowed under certain conditions. 9 Abadie and Gay (2005) included religion with a Catholic country indicator, which is based on the majority of population being either Catholic or not. 10 We use being Roman Catholic and having no religion to account for religious reasons for refusal, because these are the major two categories that are consistently defined across the surveys that we used. Copyright © 2014 John Wiley & Sons, Ltd.
Health Econ. 24: 1560–1572 (2015) DOI: 10.1002/hec
1562
Z. B. UGUR
Table I. Attitudes to organ donation by consent regime % willing to donate (own)
Yes No Do not know N
% willing to donate (family member)
Presumed
Informed
Presumed
Informed
55.83 26.99 17.18 48,608
55.56 25.45 19.00 22,494
52.18 25.08 22.73 48,608
50.81 23.18 26.01 22,494
Source: Eurobarometer 2002, 2006 and 2009 surveys.
2. METHODS This study aims to examine the impact of presumed consent on cadaveric donations and kidney transplantations using a panel dataset from the EU-27 countries plus Croatia11 in the period 2000–2010. To identify the impact of presumed consent, we would ideally need country fixed effects models, which would treat unobserved country-level heterogeneity. However, because there have been few changes in legislation over the last 20 years in Europe, we cannot estimate country fixed effects reliably. Instead, we estimate a pooled ordinary least squares (OLS) model. However, the results would be biased if presumed consent is legislated in countries where there is higher social acceptance of organ donation. To ensure the reliability of the pooled OLS estimates, we follow a three-step approach. We firstly study differences in willingness to donate organs in presumed and informed consent countries. If people in presumed consent countries were no more willing to donate organs than in informed consent countries, we would be less concerned that the pooled OLS results were driven by differences in public attitudes toward donation. Secondly, we study differences in registering preferences for organ donation in presumed and informed consent countries by looking at organ donation card holding. Because people generally fail to take action, we expect to observe lower registration of unwillingness in presumed consent countries. The third step, which forms our main analysis, explores the impact of presumed consent legislation on cadaveric donations and kidney transplantations. 2.1. The relationship between willingness to donate and presumed consent In this section, we analyze willingness to donate and willingness to give consent for a family member using individual level data from the 2002, 2006 and 2009 Eurobarometer surveys. The variable ‘willingness to donate organs’ is derived from the following question: ‘Would you be willing to donate ONE of your organs to an organ donation service immediately after your death?’ In our setting, a person who says ‘No’ is a strict nondonor, whereas a person who says ‘Yes’ may agree to donate some of her organs but may still refuse to donate other organs. Eurobarometer surveys are conducted through face to face interviews. Potentially, responses to these questions might not fully reflect true preferences for organ donation, given the respondents’ tendency to give socially desirable answers. The rest of study is based on the assumption that people from informed and presumed consent countries do not differ in their tendency to give socially desirable responses. Table I provides descriptive information on the willingness to donate indicators. There are no significant differences in terms of percentage of individuals who are willing to donate their own organs between presumed 11
We use countries for which a Eurobarometer survey was conducted in 2002, 2006 or 2009 because these surveys are the only surveys in which organ donation preferences are available. These countries/regions are listed in Appendix C. Macedonia and the Turkish Cypriot Community are eliminated because organ donation figures are not available for them. Malta and Macedonia are discarded because relevant legislation could not be found. Turkey is discarded because causes of death, which constitute the bulk of cadaveric donors, were not available.
Copyright © 2014 John Wiley & Sons, Ltd.
Health Econ. 24: 1560–1572 (2015) DOI: 10.1002/hec
1563
PRESUMED CONSENT
Table II. Pooled ordinary least squares willingness to donate regression results Willingness to donate (own) (1) a. All Presumed Baseline controls N b. Excluding Spain Presumed Baseline controls N
0.013 (0.085) – 71,102 0.021 (0.086) – 68,073
(2) 0.003 (0.073) + 70,085 0.009 (0.073) + 67,085
Willingness to donate (Family Member) (1)
(2)
0.005 (0.069) – 71,102 0.016 (0.070) – 68,073
0.008 (0.063) + 70,085 0.006 (0.063) + 67,085
Standard errors (in parentheses) are clustered at country level, *p < .1, **p < .05, ***p < .01. Baseline controls: gender, age, age, residence in an urban or small town (vs. the omitted category of large town), country, marital status, occupation, age of the respondent at which her full-time education ended, and fixed telephone line and mobile telephone ownership.
and informed consent countries. However, individuals who are willing to give consent for a family member are significantly more prevalent in presumed consent countries compared with informed consent countries. For both indicators, a large percentage of people were undecided, which implies that preferences for organ donation are not clear-cut. Therefore, presumed consent legislation could produce more organ donation especially in people who do not have a preference. Using these surveys, we estimate an OLS regression12 of willingness to donate one’s own organs and willingness to give consent for a deceased relative on an indicator variable of presumed consent in model (1). We add to these regressions a set of control variables in model (2). The control variables are gender, age, residence in an urban or small town (vs. the omitted category of large town), country, marital status, occupation, age of the respondent at which her full-time education ended (as an indicator for education), and fixed telephone line and mobile telephone ownership as a proxy for wealth.13 We also experiment with different samples. An initial regression analysis is conducted for the whole sample. We exclude Spain from the sample to see if this produced any changes, because Spain is well known for its success in organ donation (Matesanz and Miranda, 2002; Chang et al., 2003). The regression models are estimated with standard errors clustered at country level. The OLS results are displayed in Table II. Interestingly, we find negative coefficients for the relationship between willingness to donate one’s own organs, willingness to give consent for a deceased family member and presumed consent legislation for the whole sample. After the control variables are accounted for, the coefficient becomes positive yet insignificant. These findings imply that presumed consent legislation is not necessarily enacted in countries where there is wide social acceptance of organ donation. 2.2. The relationship between registering preferences and presumed consent In this section, we analyze whether individuals take action in line with their preferences for organ donation. In general, we expect individuals from informed consent countries to register their willingness and individuals from presumed consent countries to register their refusal with the donation card. Also, individuals from both presumed and informed consent countries are often allowed to register their willingness or refusal with the
12 13
We also estimated these models with ordered probit models. The results are very similar. We used fixed telephone line and mobile phone ownership as a proxy for wealth because this information was available in all three Eurobarometer surveys. The Eurobarometer 2002 survey has a household income variable for each country, whereas the Eurobarometer 2006 and 2009 surveys include ownership of a set of household goods such as television, DVD player, Internet access and car. Collecting information on ownership of household goods became popular with the Demographic Health Surveys.
Copyright © 2014 John Wiley & Sons, Ltd.
Health Econ. 24: 1560–1572 (2015) DOI: 10.1002/hec
1564
Z. B. UGUR
Table III. Percentage of organ donation card holding by consent regime and preferences for organ donation Total sample
Willing to donate
Not willing to donate
5.97 18.93 27,584
9.72 32.11 15,053
1.56 3.1 7793
Presumed Informed N
Source: Eurobarometer 2006 survey.
donation card, to reduce the likelihood of the family withholding consent.14 To check this, we use a variable on organ donation card holding from the Eurobarometer 2006 survey. This variable is derived from the following question: ‘Do you already have an organ donation card?’ On an organ donation card, one can often specify which organs one wants to donate or not. In our setting, a person who is unwilling to donate any of her organs is strictly a non-donor, whereas among people who are willing to donate, having an organ donation card can mean full or partial willingness in both presumed and informed consent countries. Table III provides descriptive information on donation card holding. Almost 6% of individuals in presumed consent countries hold a donation card, whereas 19% of individuals in informed consent countries have one. Of individuals willing to donate, 32% and 10% have a donation card in informed and presumed consent countries, respectively. Of individuals not willing to donate any organs, roughly 3% and 2% have a donation card in informed and presumed consent countries, respectively. In line with intuition, for individuals willing to donate, registering preferences with an organ donation card is less common in presumed consent countries. But, in contrast to intuition, we do not observe higher registration in presumed consent countries when individuals are not willing to donate. We firstly perform a regression of donation card ownership on the ‘presumed consent’ dummy variable in model (1), for the whole sample. We add to this regression a set of control variables in model (2). The control variables are gender, age, residence in an urban or small town (vs. the omitted category of large town), country, marital status, occupation, age of the respondent at which her full-time education ended (as an indicator for education), and fixed telephone line and mobile telephone ownership as a proxy for wealth. The same models are also run for the subsample of individuals who are willing to donate or not willing to donate. We also experiment with different samples. We exclude Spain from the sample to see whether this produces any changes. All regression models are estimated with standard errors clustered at the country level. For ease of interpreting the coefficients, marginal effects calculated from the probit estimation results are displayed in Table IV. We observe significantly lower organ donation card holding in presumed consent countries on average. In line with our expectations, we find lower donation card holding among individuals who are willing to donate their organs in presumed consent countries compared with informed consent countries. However, surprisingly, among people who are not willing to donate, we do not observe higher registration in presumed consent countries. These findings imply that presumed consent legislation is likely to increase cadaveric donation rates because unwilling individuals fail to register their preferences in presumed consent countries. There could be many reasons for this. For instance, informed consent countries might be more active in soliciting organ donation through advertising the donation card more actively. Another channel could be different registration costs in different countries.15 For the purposes of this study, it suffices to observe higher organ donation card holding for both individuals who are willing to donate and not willing to donate in informed consent countries. 14
For example, if a person did not register on a non-willing list in an informed consent country, family members might be more likely to give consent, because this signals that the person is not strongly against organ donation. 15 For instance, in the Netherlands and in the UK, both informed consent countries, online registration of preferences for organ donation is possible. But Sweden, a presumed consent country, also allows for online registration of preferences. Copyright © 2014 John Wiley & Sons, Ltd.
Health Econ. 24: 1560–1572 (2015) DOI: 10.1002/hec
1565
PRESUMED CONSENT
Table IV. Organ donation card holding regression results Total sample
a. All countries Presumed Baseline controls N b. Excluding Spain Presumed Baseline controls N
Willing to donate
Not willing to donate
(1)
(2)
(1)
(2)
(1)
(2)
0.130** (0.057) – 27,584
0.118** (0.047) + 27,167
0.224*** (0.070) – 15,053
0.214*** (0.062) + 14,880
0.015 (0.019) – 7793
0.012 (0.014) + 7643
0.130** (0.057) – 26,558
0.119** (0.047) + 26,167
0.224*** (0.071) – 14,487
0.215*** (0.063) + 14,333
0.015 (0.019) – 7610
0.012 (0.014) + 7464
Standard errors (in parentheses) are clustered at country level, *p < .1, **p < .05, ***p < .01. Baseline controls: gender, age, age, residence in an urban or small town (vs. the omitted category of large town), country, marital status, occupation, age of the respondent at which her full-time education ended, and fixed telephone line and mobile telephone ownership.
2.3. The impact of presumed consent on cadaveric donors and transplantations 2.3.1. Descriptive statistics. Figures 1 and 2 show higher cadaveric donor and kidney transplantation rates in presumed consent countries in comparison with informed consent countries. Yet, Bulgaria, having the lowest levels of cadaveric donors and kidney transplantations, is a presumed consent country. This suggests that there are other factors that need to be taken into account. Mainly following previous literature, information on a number of factors, such as number of deaths by specific causes, health spending, medical infrastructure, religious beliefs and education level, is gathered from a variety of sources.16 In most cases, deceased donors were brain-dead and their hearts were artificially functioning with the help of ventilation machines. The most common causes of brain death are homicides, motor vehicle accidents and cerebro-vascular diseases. According to the Eurobarometer 2009 survey, religious reasons, distrust in the system and fear of manipulation of the human body are three major causes of refusal to donate organs. To capture trust in the system, corruption perception scores from Transparency International are included. Corruption perception scores range between 10 (highly clean) and 0 (highly corrupt). To control for religiosity changes over time, we compiled percentage of population being Roman Catholic and having no religion17 from four surveys: the International Social Survey Program, the European Social Survey, the European Values Survey and Eurobarometers conducted between 1999/2000 and 2010. Table V provides summary statistics for the sample. The second column shows presumed consent countries having 16.95 cadaveric donors pmp per year, whereas the third column shows informed consent countries having 13.17 cadaveric donors pmp per year. From the fourth column, we observe presumed consent countries have 3.8 more cadaveric donors pmp per year, and this difference is statistically significant. We also observe significantly higher cadaveric donation after brain death and kidney transplantation rates in presumed consent countries. There are other differences between these countries. In particular, deaths from homicide, motor vehicle accidents and cerebro-vascular diseases are higher in presumed consent countries. However, only the motor vehicle accident difference is statistically significant. Health expenditure per capita is relatively lower, whereas the number of hospital beds per capita is relatively higher in presumed consent countries. Informed consent countries, having higher corruption perception scores, are perceived as more transparent. No major difference is observed in the percentage of the population considering themselves as having no religion, while we observe 16 17
Detailed information on sources of the data is presented in Appendix D. Includes people who consider themselves as agnostic, atheist or having no religion.
Copyright © 2014 John Wiley & Sons, Ltd.
Health Econ. 24: 1560–1572 (2015) DOI: 10.1002/hec
1566
Z. B. UGUR
Figure 1. Cadaveric organ donor rate per million population in 2010
Figure 2. Kidney transplantation from deceased donors rate per million population in 2010
a higher percentage of the population considering themselves as Roman Catholic in presumed consent countries. Relative to informed consent countries, presumed consent countries have a statistically lower percentage of upper secondary or tertiary educated male population. No significant difference was found for the upper secondary or tertiary educational attainment of females. 2.3.2. Regression output for cadaveric donation rate. In this section, we provide a regression analysis for total cadaveric donation and brain death donation rate. We differentiate between donation after brain death (known as heart-beating donation) and donation after circulatory death (known as non-heart-beating donation) because the latter is often less efficient for transplantation (Cota et al., 2013). Table VI shows the regression output for the log of total cadaveric donor rate. Model (1) contains only a dummy variable for presumed consent legislation. In model (2), to address potential confounding factors, we Copyright © 2014 John Wiley & Sons, Ltd.
Health Econ. 24: 1560–1572 (2015) DOI: 10.1002/hec
1567
PRESUMED CONSENT
Table V. Descriptive statistics (means and standard deviations for 2000–2010)
Presumed consent country Cadaveric donor rate, pmp Cadaveric donation after brain death, pmp Kidney transplant rate, pmp Homicide, pmp Deaths from motor vehicle accident rate, pmp Deaths from cerebro-vascular diseases, pmp Health expenditure per capita Hospital beds, per 100,000 population Corruption perception score % of people having no religion % of people being Roman Catholic % of higher educational attainment, males % of higher educational attainment, females Number of countries
Entire sample
Presumed consent
Informed consent
Difference
0.74 [0.44] 15.96 [7.65] 15.40 [7.36] 25.03 [11.38] 21.48 [25.80] 113.32 [48.99] 853.95 [498.73] 2238.78 [1260.52] 580.83 [166.77] 6.28 [1.99] 0.23 [0.18] 0.42 [0.35] 68.62 [13.13] 67.49 [12.07] 27
16.95 [7.96] 16.58 [7.65] 26.54 [11.75] 21.64 [25.96] 119.08 [42.60] 867.44 [473.35] 2144.57 [1236.08] 582.62 [160.31] 5.98 [1.84] 0.23 [0.18] 0.45 [0.35] 67.45 [14.62] 66.70 [13.38] 20
13.17 [5.88] 12.09 [5.24] 20.68 [8.94] 21.01 [25.48] 97.47 [60.91] 816.88 [564.44] 2507.97 [1298.67] 575.72 [185.03] 7.12 [2.16] 0.22 [0.17] 0.34 [0.34] 71.95 [6.39] 69.73 [6.74] 7
3.78*** (1.00) 4.49*** (0.95) 5.85*** (1.48) 0.63 (3.45) 21.61*** (6.45) 50.55 (66.86) 363.40* (165.84) 6.90 (22.26) 1.14*** (0.26) 0.01 (0.02) 0.12* (0.05) 4.50** (1.72) 3.03 (1.59)
Notes: standard deviations in [ ]; standard errors in ( ). *p < 0.05, **p < 0.01, ***p < 0.001.
Table VI. Pooled ordinary least squares estimates of log cadaveric donor rate (1) Legislation Presumed consent Practicing legislation Family consent Presumed consent * family consent Potential donors Log of MVA + CVD + homicide, pmp Health spending Log of health expenditure per capita Medical infrastructure Log of hospital beds per 100,000 people Trust in the system Corruption perception score Religious beliefs % of people having no religion % of people being Roman Catholic Education % of higher educational attainment, males % of higher educational attainment, females Average willingness to donate Country group fixed effects R-squared N
0.355 (0.401)
(2) 0.329** (0.157)
(3) 0.390** (0.148)
(4) 0.350* (0.190)
(5) 0.042 (0.163) 0.309 (0.341) 0.552* (0.291)
– 0.037 287
0.038 (0.076)
0.072 (0.076)
0.108 (0.072)
0.072 (0.068)
0.227 (0.181)
0.212 (0.185)
0.190 (0.212)
0.190 (0.162)
0.073 (0.273)
0.226 (0.258)
0.092 (0.280)
0.151 (0.241)
0.117*** (0.042)
0.160*** (0.048)
0.079 (0.059)
0.071 (0.054)
1.824*** (0.345) 1.237*** (0.254)
1.818*** (0.347) 1.278*** (0.255)
1.914*** (0.508) 1.353*** (0.355)
1.873*** (0.487) 1.381*** (0.306)
0.038** (0.017)
0.040** (0.015)
0.007 (0.020)
0.000 (0.019)
0.043** (0.018)
0.045** (0.017)
0.007 (0.025)
0.001 (0.022)
– 0.655 262
1.153 (0.809) – 0.664 262
+ 0.741 262
+ 0.755 262
Standard errors (in parentheses) are clustered at country level, *p < .1, **p < .05, ***p < .01.
include control variables such as death rates from homicides, motor vehicle accidents and cerebro-vascular diseases, health expenditure per capita, hospital beds per 100,000 population, corruption perception scores, percentage of population considering themselves as having no religion and Roman Catholic, and Copyright © 2014 John Wiley & Sons, Ltd.
Health Econ. 24: 1560–1572 (2015) DOI: 10.1002/hec
1568
Z. B. UGUR
percentage of upper secondary or tertiary educated males and females. According to model (2), the coefficient of the presumed consent indicates a 33% higher cadaveric donor rate. This difference is significant at the 5% level. Model (3) includes average willingness to donate organs in each country from the Eurobarometer surveys conducted in 2002, 2006 and 2009, to address the potential confounding effect of social acceptance of organ donation. The coefficient of presumed consent dummy is still significant, suggesting that the impact of presumed consent on the cadaveric donor rate cannot be attributed to higher social acceptance of organ donation. To capture the idea that presumed consent might be enacted in countries where there is different unobserved heterogeneity, we group countries in model (4) according to the following classification, with presumed consent countries in bold: (1) (2) (3) (4) (5) (6) (7) (8) (9)
Ireland–UK Greece–Cyprus Germany–Austria–Hungary (Hungary is included here because of the historical connection between Hungary and Austria) Netherlands–Belgium–Luxembourg Estonia–Lithuania–Latvia (These countries together formed the Balttransplant organization) Denmark–Sweden–Finland (which formed the Scandiatransplant organization) Poland–Czech Republic–Slovakia (Western Slavic ethnic origin) Portugal–Spain–Italy–France (Latin ethnic origin) Bulgaria–Romania–Croatia–Slovenia (Southern-Slavic ethnic origin. These countries are also known as Balkan countries)
Model (4) shows 35% higher donation rates in presumed consent countries when country group dummy variables are included in the model. In model (5), we included family consent and its interaction with presumed consent to check whether family consent makes any difference. Model (5) suggests that presumed consent does matter especially when combined with family consent, whereas in informed consent countries, seeking family consent has a negative but insignificant coefficient. This result is in line with the prediction that even if families make the final decision on organ donation, presumed consent can lead to notably higher donation rates. The same five models of the cadaveric donor rate are also estimated for the log of brain death donor rate. Table VII provides the regression output. Overall, the regression results suggest a more pronounced impact of presumed consent.
2.3.3. Regression output for kidney transplantation rate. In this section, we provide a regression analysis for the kidney transplantation rate. We use the same five models of the cadaveric donor rate to analyze the log kidney transplant rate. Table VIII provides the regression results. Model (1) shows that presumed consent countries have a 33% higher kidney transplant rate. Similar to cadaveric donor regression results, models (2) and (3) report that presumed consent countries have statistically significantly higher kidney transplant rates. In model (4), when country group dummy variables are included, presumed consent countries are found to have a 31% higher kidney transplant rate. A similar pattern emerges from model (5) as with the cadaveric donor rate. Presumed consent increases the kidney transplant rate when combined with family consent. In line with intuition, family consent in informed countries has a negative yet insignificant coefficient. Overall, the regression results from Tables VI and VIII suggest that presumed consent countries have 28% to 32% higher cadaveric donation and 27% to 31% higher kidney transplant rates than informed consent countries. We conducted a number of robustness checks, such as inclusion of year fixed effects and exclusion of Spain from the sample. The weighted least squares regression is also estimated, in which weighting proportional to size of population is used. Details of these tests are reported in Appendix E. Copyright © 2014 John Wiley & Sons, Ltd.
Health Econ. 24: 1560–1572 (2015) DOI: 10.1002/hec
1569
PRESUMED CONSENT
Table VII. Pooled ordinary least squares estimates of log of brain death donor rate
Legislation Presumed consent Practicing legislation Family consent * Presumed consent family consent Baseline controls Country group fixed effects R-squared N
(1)
(2)
(3)
(4)
(5)
0.408 (0.388)
0.386** (0.151)
0.422*** (0.144)
0.366** (0.161)
0.005 (0.136) 0.372 (0.279) 0.647** (0.245)
– – 0.050 287
+ – 0.665 262
+ – 0.668 262
+ + 0.762 262
+ + 0.781 262
Standard errors (in parentheses) are clustered at country level, *p < .1, **p < .05, ***p < .01. Baseline controls: death rates from homicides, motor vehicle accidents and cerebro-vascular diseases, health expenditure per capita, hospital beds per 100,000 population, corruption perception scores, percentage of population considering themselves as having no religion and Roman Catholic, and percentage of upper secondary or tertiary educated males and females.
Table VIII. Pooled ordinary least squares estimates of log kidney transplant rates
Legislation Presumed consent Practicing legislation Family consent * Presumed consent Family consent Baseline controls Country group fixed effects R-squared N
(1)
(2)
(3)
(4)
(5)
0.328 (0.387)
0.347** (0.154)
0.367** (0.153)
0.309* (0.178)
0.015 (0.141) 0.180 (0.315) 0.525* (0.278)
– – 0.030 296
+ – 0.614 270
+ – 0.615 270
+ + 0.737 270
+ + 0.756 270
Standard errors (in parentheses) are clustered at country level, *p < .1, **p < .05, ***p < .01. Baseline controls: death rates from homicides, motor vehicle accidents and cerebro-vascular diseases, health expenditure per capita, hospital beds per 100,000 population, corruption perception scores, percentage of population considering themselves as having no religion and Roman Catholic, and percentage of upper secondary or tertiary educated males and females.
3. DISCUSSION In this section, firstly, we discuss further whether presumed consent is an indicator of a country’s commitment to organ donation rather than a causal mechanism in itself. Secondly, we discuss the advantages and disadvantages of both types of consent legislation. A country’s commitment to organ donation could be indicated by available transplant capacity. In our main analysis, transplantation capacity is imperfectly proxied by the number of hospital beds per 100,000 people. Models (1) and (2) in Table IX use the same specifications as models (3) and (4) in Table VI, with the logarithm of number of transplantation centers pmp as an additional control variable. Table IX reports an increased coefficient of presumed consent compared with Table VI. Kidney transplant capacity (defined as the number of kidney transplant centers pmp) could also influence kidney transplantation rates. To check this idea, models (3) and (4) in Table IX use the same specifications as models (3) and (4) in Table VIII, with the logarithm of number of transplantation centers pmp as an additional control variable. Models (3) and (4) also report an increased coefficient of presumed consent compared with Table VIII. Copyright © 2014 John Wiley & Sons, Ltd.
Health Econ. 24: 1560–1572 (2015) DOI: 10.1002/hec
1570
Z. B. UGUR
Table IX. Transplant infrastructure and presumed consent (pooled ordinary least squares) Log cadaveric donor rate
Legislation Presumed consent Transplant Infrastructure Log of kidney transplant centers Baseline controls Average willingness to donate Country group fixed effects R-squared N
Log kidney transplant rate
(1)
(2)
(3)
(4)
0.546*** (0.147)
0.683*** (0.164)
0.494*** (0.139)
0.561*** (0.161)
0.299** (0.128) + + – 0.778 153
0.274 (0.172) + – + 0.828 153
0.352** (0.141) + + – 0.772 153
0.322** (0.153) + – + 0.836 153
Standard errors (in parentheses) are clustered at country level, *p < .1, **p < .05, ***p < .01. Baseline controls: death rates from homicides, motor vehicle accidents and cerebro-vascular diseases, health expenditure per capita, hospital beds per 100,000 population, corruption perception scores, percentage of population considering themselves as having no religion and Roman Catholic, and percentage of upper secondary or tertiary educated males and females.
On the whole, the results of Table IX suggest that the regression results provided in Tables VI and VIII are not confounded by commitment to organ donation differences across countries. As regards disadvantages, both presumed consent and informed consent legislation are error-prone. In presumed consent legislation, non-willing individuals’ organs could be mistakenly removed if they did not register their non-willingness (Gill, 2004; Orentlicher, 2008). In this case, removal of an organ is wrong because it is against the patient’s wishes. In informed consent countries, willing individuals’ organs could be mistakenly not removed if they did not register their willingness (Gill, 2004; Orentlicher, 2008). In this case, non-removal of an organ is a waste of a scarce resource that could be used for improving the life of another person. So, in both scenarios, there is a possibility that some people’s wishes may not be respected. Using descriptive information from Tables I and III, we can calculate which regime produces the least percentage of errors and which regime maximizes the percentage of people whose wishes are respected. The probability of making a mistake in presumed consent legislation is 43%, which is calculated by multiplying the percentage of non-willing individuals with the probability of non-registering their preference (= (100 55.83) * (100 1.56) / 100). The probability of making a mistake in informed consent legislation is 37.71, which is given by multiplying the percentage of willing individuals with the probability of nonregistering their preference (= 55.56 * (100 32.11) / 100). If a default consent regime should be chosen such that the least number of errors is made, an informed consent regime seems to be superior. The percentage of people whose wishes are respected in an informed consent regime is those who are not willing to donate (25.45%) plus those who are willing to donate and registered their preference (= 55.56 * 32.11 / 100). This corresponds to 43.29%. The percentage of people whose wishes are respected in a presumed consent regime is those who are willing to donate (55.83%) plus those who are not willing to donate and registered their preference (= 26.99 * 1.56 / 100). This corresponds to 56.25%. If the criterion for choosing the default consent regime is to maximize the percentage of people whose wishes are respected, presumed consent seems to outperform the informed consent regime. As regards advantages, presumed consent may easily outperform informed consent legislation in terms of administration costs. Although we do not have data about how much it costs to administer presumed consent and informed consent legislation, it is highly probable that an informed consent regime is more costly because the organ donation authority of an informed consent country has to invest sizeable amounts of money in convincing people to be organ donors. Moreover, these investments have a very low probability of return, because the likelihood of dying under conditions which would be suitable for organs to be transplanted is very low. Howard and Byrne (2007) estimated the probability of a potential donor being an actual donor at some point in her lifetime to be 0.0028. Copyright © 2014 John Wiley & Sons, Ltd.
Health Econ. 24: 1560–1572 (2015) DOI: 10.1002/hec
PRESUMED CONSENT
1571
In brief, if policy makers attribute greater importance to minimizing the number of errors, informed consent seems to be preferable. If policy makers attribute greater importance to saving lives, respecting people’s wishes or minimizing administration costs, presumed consent seems to be preferable.
4. LIMITATIONS Currently, the Eurobarometer surveys are the only source that allows cross-country comparisons on organ donation preferences. However, they do have some limitations. Firstly, the organ donation preferences are based on self-reporting by survey respondents. The quality of the information may therefore be affected by over-reporting due to social desirability. Our analysis in this study is based on the assumption that people in presumed and informed consent countries over-report their willingness to donate by the same amount. However, preference for organ donation in a country is potentially affected by the consent law because this sends a signal about the social norm on organ donation. Therefore, measures of willingness to donate will appear higher for people from presumed consent countries than informed consent countries. In this case, our results on which consent regime produces the least percentage of errors might be misleading. Secondly, the question about organ donation preference is asked in terms of whether the respondent wants to donate one of her organs or not. A person may very well agree to donate some of her organs but not agree to donate other organs. Unfortunately, the dataset does not allow tracking of organ donation preferences for specific organs. In reality, organ donation cards often let people declare which organs they want to donate or not. However, the organ donation card question is formulated as a 0/1 type question, making it impossible to comment on partial donation registration. Thirdly, the wording of the organ donation card question might lead respondents to think that having the card only refers to people who have declared that they want to be donors. A better wording could be, ‘Do you have an organ donation declaration or refusal card/ registration?’
5. CONCLUSION In this study, we attempt to extend the literature on how presumed consent impacts cadaveric donors and kidney transplantations in the EU-27 countries plus Croatia in the period 2000–2010. As a first step, we show evidence that presumed consent is not necessarily legislated in countries where there is higher social acceptance of organ donation. In the second step, we show that people fail to take action in line with their preferences, which is especially relevant for individuals who are unwilling to donate their organs. Therefore, we suggest that presumed consent is likely to produce more organ donors. In the main analysis, after accounting for potential confounding factors, our estimates suggest that presumed consent countries have 28% to 32% higher cadaveric donation and 27% to 31% higher kidney transplant rates than informed consent countries. Our study indicates that presumed consent legislation can be instrumental in saving lives, respecting people’s wishes and minimizing the administration costs.
CONFLICT OF INTEREST The authors have no conflict of interest.
ACKNOWLEDGEMENTS
I thank Peter Kooreman, Shelly Lundberg, Jan van Ours, Marcel Das, Matthijs Kalmijn, seminar participants at Tilburg University and Lowland Health Economists’ Study Group (LolaHESG). Copyright © 2014 John Wiley & Sons, Ltd.
Health Econ. 24: 1560–1572 (2015) DOI: 10.1002/hec
1572
Z. B. UGUR
REFERENCES
Abadie A, Gay S. 2005. The impact of consent legislation on cadeveric organ donation: a cross country study. NBER Working Paper No. 10604. Becker G, Elías J. 2007. Introducing incentives in the market for live and cadaveric organ donations. The Journal of Economic Perspectives 21: 3–24. Bilgel F. 2010. The impact of presumed consent laws and institutions on deceased organ donation. The European Journal of Health Economics 13: 29–38. Chang G, Mahanty H, Ascher N, Roberts J. 2003. Expanding the donor pool: can the Spanish model work in the United States? American Journal of Transplantation 3: 1259–1263. Cohen LR. 1989. Increasing the supply of transplant organs: the virtues of a futures market. George Washington Law Review 58: 1–51. Cota N, Burgess M, English W. 2013. Organ donation after circulatory death. Anaesthesia Tutorial of the Week 282: 2. Davis CL, Delmonico FL. 2005. Living-donor kidney transplantation: a review of the current practices for the live donor. Journal of American Society of Nephrology 16: 2098–2110. Ellison M, McBride M, Taranto S, Delmonico F. 2002. Living kidney donors in need of kidney transplants: a report from the Organ Procurement and Transplantation Network. Transplantation 74: 1349–1351. Gevers S, Janssen A, Friele R. 2004. Consent systems for post mortem organ donation in Europe. European Journal of Health Law 11: 175–186. Gill M. 2004. Presumed consent, autonomy, and organ donation. Journal of Medicine and Philosophy 29: 37–59. Global Observatory on Donation and Transplantation. 2011, February 9. Retrieved from http://data.transplant-observatory. org/paginas/informes/DatosUsuario.aspx Gnant M, Wamser P, Goetzinger P, Sautner T, Steininger R, Muehlbacher F. 1991. The impact of the presumed consent law and a decentralized organ procurement system on organ donation: quadruplication in the number of organ donors. Transplant Proceedings 23: 2685–2686. Greene W. 2011. Fixed Effects vector decomposition: a magical solution to the problem of time invariant variables in fixed effects models? Political Analysis 19: 135–146. Healy K. 2005. The political economy of presumed consent. Working Paper 31, University of California, Los Angeles. Howard D. 2007. Producing organ donors. The Journal of Economic Perspectives 21: 25–36. Howard DH, Byrne MM. 2007. Should we promote organ donor registries when so few registrants will end up being donors? Medical Decision Making 27: 243–249. Johnson E, Goldstein D. 2003. Do defaults save lives? Science 302: 1338–1339. Matesanz R, Miranda B. 2002. A decade of continuous improvement in cadaveric organ donation: the Spanish model. Journal of Nephrology 15: 22–28. Newsletter Transplant. 2009. http://www.ont.es/publicaciones/Paginas/Publicaciones.aspx. (accessed February 11, 2011) Newsletter Transplant. 2010. http://www.ont.es/publicaciones/Paginas/Publicaciones.aspx. (accessed February 11, 2011) Orentlicher D. 2008. Presumed consent to organ donation: its rise and fall in the United States. Rutgers Law Review 61: 295–331. Oz M, Kherani A, Rowe A, Roels A, Crandall C, Tomatis L, et al. 2003. How to improve organ donation: results of the ISHLT/FACT poll. Journal of Heart and Lung Transplantation 22: 389–410. Paparde I. 2010. Latvian public unwilling to donate organs. Retrieved 9 29, 2012, from http://ec.europa.eu/health/ blood_tissues_organs/docs/art_latvia_2010_en.pdf Roels L, Vanrenterghem Y, Waer M, Christiaens M, Gruwez J, Michielsen P. 1991. 3 years of experience with a presumed consent legislation in Belgium—its impact on multiorgan donation in comparison with other European countries. Transplantation Proceedings 23: 903–904. Roth A. 2007. Repugnance as a constraint on markets repugnance as a constraint on markets. The Journal of Economic Perspectives 21: 37–58.
SUPPORTING INFORMATION Additional supporting information may be found in the online version of this article at the publisher's web site.
Copyright © 2014 John Wiley & Sons, Ltd.
Health Econ. 24: 1560–1572 (2015) DOI: 10.1002/hec
DOES PRESUMED CONSENT SAVE LIVES? EVIDENCE FROM EUROPE ZEYNEP BURCU UGUR* Department of Economics, Tilburg University, Tilburg, Netherlands
SUMMARY One policy tool that could affect organ donation rates is legislative defaults. In this study, we examine how presumed consent impacts cadaveric donations and kidney transplantations, using a panel dataset from the EU-27 countries plus Croatia in the period 2000–2010. We find that presumed consent countries have 28% to 32% higher cadaveric donation and 27% to 31% higher kidney transplant rates in comparison to informed consent countries, after accounting for potential confounding factors. After studying willingness to donate one’s organs and registering preferences for organ donation, we find that presumed consent could increase cadaveric donation rates, because people fail to register their preferences and many have no preference for organ donation. Copyright © 2014 John Wiley & Sons, Ltd. Received 22 January 2014; Revised 10 August 2014; Accepted 8 September 2014 JEL Classification: KEY WORDS:
I18; K32; D82
presumed consent legislation; organ donation; kidney transplantations
1. INTRODUCTION The question of how to generate more organ donors is highly relevant, because the chronic shortage1 of organs is directly linked to the death of many patients, both in Europe and elsewhere. One policy option is to adjust legislative defaults. There are two types of legislation for obtaining consent for organ transplantation: informed consent and presumed consent. In informed consent, individuals declare their willingness explicitly by registering to be an organ donor. In presumed consent, a brain-dead individual whose organs are suitable for transplantation is automatically considered to be a donor unless she has stated a preference for not donating.2 Theoretically, if the costs of registering preferences for organ donation are low,3 the defaults would not have a great effect for fully rational individuals who already have established preferences for organ donation. In the case of a mismatch between the default and the preferences, individuals are expected to take action for the desired option. However, if individuals are more likely to accept the effortless default option rather than make a choice that has mental costs, defaults might matter. The mental costs of making a decision for organ donation could be substantial, because it requires thinking about one’s own death, which is generally perceived as *Correspondence to: Department of Economics C214, Hukumet Meydani No: 2 Ulus, Social Sciences University of Ankara, Ankara, Turkey. E-mail: [email protected] 1
See Appendix A for more information on the extent of the organ shortage problem in Europe. In practice, in some presumed and informed consent countries, consent from the family of the deceased is routinely sought even if he or she explicitly stated their preference to be a donor. 3 There may be differences in registration costs between countries: some countries allow for registering organ donation preferences through online forms, others have dedicated telephone lines. Some only accept applications by mail. Some only accept registration at specific locations, which may require incurring transportation costs. 2
Copyright © 2014 John Wiley & Sons, Ltd.
PRESUMED CONSENT
1561
unpleasant and stressful. Thus, there might be differences in terms of the age structure of the donor pool between presumed and informed consent systems. Young people, whose organs are more valuable for transplantation, are less likely to think about death and register as organ donors, because they may anticipate a long life. Presumed consent could solve the registration problem for young people. Even if presumed consent is not strictly enforced, in that family consent is always sought, legislating for presumed consent might still be a positive signal from the government to the families. Presumed consent legislation might even impact the way in which doctors talk to families. Doctors then have the law on their side when trying to explain the need for organ donation to a bereaved family. Some studies find higher organ donation rates after the enactment of presumed consent legislation. After the introduction of presumed consent in Austria in 1982, the donor rate quadrupled (Gnant et al., 1991). Similarly, Roels et al. (1991) reported more than a doubling of kidney donations in Belgium after the introduction of presumed consent in 1986.4,5 Some studies (Johnson and Goldstein, 2003; Abadie and Gay, 2005; Bilgel, 2010) find higher organ donation rates in presumed consent countries compared with informed consent countries. However, the evidence is not entirely convincing. For instance, Abadie and Gay (2005) acknowledged that if presumed consent is enacted in countries where there is higher social acceptance of organ donation, the found effect might be biased. Bilgel (2010) used the fixed effects vector decomposition, which is not econometrically valid (Greene, 2011). Moreover, there is no consensus in the literature on whether the presumed consent legislation is an indicator of a country’s commitment to organ donation or a causal mechanism in itself (Healy, 2005). Yet, according to medical professional opinion,6 the most effective option for alleviating organ shortage is enactment of presumed consent (Oz et al., 2003). The other potential channels7 do prompt potential donors to take action but do not significantly influence donation decisions in the population as a whole. This study contributes to the literature in some important ways. We firstly address potential endogeneity of presumed consent by showing evidence that presumed consent is not necessarily legislated in countries where there is higher public support for organ donation. Secondly, we address the claim that presumed consent is an indicator of a country’s commitment to organ donation rather than a causal mechanism in itself. We show that after taking into account a country’s commitment to organ donation, proxied by kidney transplant centers as an additional control variable, the coefficient of presumed consent is still statistically significant and even increases. Thirdly, although according to the Eurobarometer 2009 survey, religious reasons,8 distrust in the system and fear of manipulation of the human body are three major causes of refusing organ donation, previous studies have not dealt adequately with trust in the system and with religious differences.9 To capture trust in the system, we included corruption perception scores from Transparency International. To control for religiosity changes over time, we compiled the percentage of people being Roman Catholic10 and having no religion, from various surveys. Lastly, to the author’s knowledge, this study is the first to analyze the impact of presumed consent on kidney transplantation, which is of particular relevance from a policy perspective.
4
Legislative change in Austria and Belgium is not used in this study, because these changes occurred before international country-level organ donation data were available. 5 The International Registry of Organ Donation and Transplantation provides organ donation rates from 1993 and transplantation rates from 1999. 6 The results are based on a survey of 739 members of the International Society for Heart and Lung Transplantation on improving organ donation levels. 7 See Appendix B for discussion on other policy options for promoting organ donation. 8 Although the Roman Catholic Church explicitly supports organ donation, some denominations such as Jehovah’s Witnesses and the Christian Scientists discourage these practices. In Islam and Judaism, organ donations and transplants are allowed under certain conditions. 9 Abadie and Gay (2005) included religion with a Catholic country indicator, which is based on the majority of population being either Catholic or not. 10 We use being Roman Catholic and having no religion to account for religious reasons for refusal, because these are the major two categories that are consistently defined across the surveys that we used. Copyright © 2014 John Wiley & Sons, Ltd.
Health Econ. 24: 1560–1572 (2015) DOI: 10.1002/hec
1562
Z. B. UGUR
Table I. Attitudes to organ donation by consent regime % willing to donate (own)
Yes No Do not know N
% willing to donate (family member)
Presumed
Informed
Presumed
Informed
55.83 26.99 17.18 48,608
55.56 25.45 19.00 22,494
52.18 25.08 22.73 48,608
50.81 23.18 26.01 22,494
Source: Eurobarometer 2002, 2006 and 2009 surveys.
2. METHODS This study aims to examine the impact of presumed consent on cadaveric donations and kidney transplantations using a panel dataset from the EU-27 countries plus Croatia11 in the period 2000–2010. To identify the impact of presumed consent, we would ideally need country fixed effects models, which would treat unobserved country-level heterogeneity. However, because there have been few changes in legislation over the last 20 years in Europe, we cannot estimate country fixed effects reliably. Instead, we estimate a pooled ordinary least squares (OLS) model. However, the results would be biased if presumed consent is legislated in countries where there is higher social acceptance of organ donation. To ensure the reliability of the pooled OLS estimates, we follow a three-step approach. We firstly study differences in willingness to donate organs in presumed and informed consent countries. If people in presumed consent countries were no more willing to donate organs than in informed consent countries, we would be less concerned that the pooled OLS results were driven by differences in public attitudes toward donation. Secondly, we study differences in registering preferences for organ donation in presumed and informed consent countries by looking at organ donation card holding. Because people generally fail to take action, we expect to observe lower registration of unwillingness in presumed consent countries. The third step, which forms our main analysis, explores the impact of presumed consent legislation on cadaveric donations and kidney transplantations. 2.1. The relationship between willingness to donate and presumed consent In this section, we analyze willingness to donate and willingness to give consent for a family member using individual level data from the 2002, 2006 and 2009 Eurobarometer surveys. The variable ‘willingness to donate organs’ is derived from the following question: ‘Would you be willing to donate ONE of your organs to an organ donation service immediately after your death?’ In our setting, a person who says ‘No’ is a strict nondonor, whereas a person who says ‘Yes’ may agree to donate some of her organs but may still refuse to donate other organs. Eurobarometer surveys are conducted through face to face interviews. Potentially, responses to these questions might not fully reflect true preferences for organ donation, given the respondents’ tendency to give socially desirable answers. The rest of study is based on the assumption that people from informed and presumed consent countries do not differ in their tendency to give socially desirable responses. Table I provides descriptive information on the willingness to donate indicators. There are no significant differences in terms of percentage of individuals who are willing to donate their own organs between presumed 11
We use countries for which a Eurobarometer survey was conducted in 2002, 2006 or 2009 because these surveys are the only surveys in which organ donation preferences are available. These countries/regions are listed in Appendix C. Macedonia and the Turkish Cypriot Community are eliminated because organ donation figures are not available for them. Malta and Macedonia are discarded because relevant legislation could not be found. Turkey is discarded because causes of death, which constitute the bulk of cadaveric donors, were not available.
Copyright © 2014 John Wiley & Sons, Ltd.
Health Econ. 24: 1560–1572 (2015) DOI: 10.1002/hec
1563
PRESUMED CONSENT
Table II. Pooled ordinary least squares willingness to donate regression results Willingness to donate (own) (1) a. All Presumed Baseline controls N b. Excluding Spain Presumed Baseline controls N
0.013 (0.085) – 71,102 0.021 (0.086) – 68,073
(2) 0.003 (0.073) + 70,085 0.009 (0.073) + 67,085
Willingness to donate (Family Member) (1)
(2)
0.005 (0.069) – 71,102 0.016 (0.070) – 68,073
0.008 (0.063) + 70,085 0.006 (0.063) + 67,085
Standard errors (in parentheses) are clustered at country level, *p < .1, **p < .05, ***p < .01. Baseline controls: gender, age, age, residence in an urban or small town (vs. the omitted category of large town), country, marital status, occupation, age of the respondent at which her full-time education ended, and fixed telephone line and mobile telephone ownership.
and informed consent countries. However, individuals who are willing to give consent for a family member are significantly more prevalent in presumed consent countries compared with informed consent countries. For both indicators, a large percentage of people were undecided, which implies that preferences for organ donation are not clear-cut. Therefore, presumed consent legislation could produce more organ donation especially in people who do not have a preference. Using these surveys, we estimate an OLS regression12 of willingness to donate one’s own organs and willingness to give consent for a deceased relative on an indicator variable of presumed consent in model (1). We add to these regressions a set of control variables in model (2). The control variables are gender, age, residence in an urban or small town (vs. the omitted category of large town), country, marital status, occupation, age of the respondent at which her full-time education ended (as an indicator for education), and fixed telephone line and mobile telephone ownership as a proxy for wealth.13 We also experiment with different samples. An initial regression analysis is conducted for the whole sample. We exclude Spain from the sample to see if this produced any changes, because Spain is well known for its success in organ donation (Matesanz and Miranda, 2002; Chang et al., 2003). The regression models are estimated with standard errors clustered at country level. The OLS results are displayed in Table II. Interestingly, we find negative coefficients for the relationship between willingness to donate one’s own organs, willingness to give consent for a deceased family member and presumed consent legislation for the whole sample. After the control variables are accounted for, the coefficient becomes positive yet insignificant. These findings imply that presumed consent legislation is not necessarily enacted in countries where there is wide social acceptance of organ donation. 2.2. The relationship between registering preferences and presumed consent In this section, we analyze whether individuals take action in line with their preferences for organ donation. In general, we expect individuals from informed consent countries to register their willingness and individuals from presumed consent countries to register their refusal with the donation card. Also, individuals from both presumed and informed consent countries are often allowed to register their willingness or refusal with the
12 13
We also estimated these models with ordered probit models. The results are very similar. We used fixed telephone line and mobile phone ownership as a proxy for wealth because this information was available in all three Eurobarometer surveys. The Eurobarometer 2002 survey has a household income variable for each country, whereas the Eurobarometer 2006 and 2009 surveys include ownership of a set of household goods such as television, DVD player, Internet access and car. Collecting information on ownership of household goods became popular with the Demographic Health Surveys.
Copyright © 2014 John Wiley & Sons, Ltd.
Health Econ. 24: 1560–1572 (2015) DOI: 10.1002/hec
1564
Z. B. UGUR
Table III. Percentage of organ donation card holding by consent regime and preferences for organ donation Total sample
Willing to donate
Not willing to donate
5.97 18.93 27,584
9.72 32.11 15,053
1.56 3.1 7793
Presumed Informed N
Source: Eurobarometer 2006 survey.
donation card, to reduce the likelihood of the family withholding consent.14 To check this, we use a variable on organ donation card holding from the Eurobarometer 2006 survey. This variable is derived from the following question: ‘Do you already have an organ donation card?’ On an organ donation card, one can often specify which organs one wants to donate or not. In our setting, a person who is unwilling to donate any of her organs is strictly a non-donor, whereas among people who are willing to donate, having an organ donation card can mean full or partial willingness in both presumed and informed consent countries. Table III provides descriptive information on donation card holding. Almost 6% of individuals in presumed consent countries hold a donation card, whereas 19% of individuals in informed consent countries have one. Of individuals willing to donate, 32% and 10% have a donation card in informed and presumed consent countries, respectively. Of individuals not willing to donate any organs, roughly 3% and 2% have a donation card in informed and presumed consent countries, respectively. In line with intuition, for individuals willing to donate, registering preferences with an organ donation card is less common in presumed consent countries. But, in contrast to intuition, we do not observe higher registration in presumed consent countries when individuals are not willing to donate. We firstly perform a regression of donation card ownership on the ‘presumed consent’ dummy variable in model (1), for the whole sample. We add to this regression a set of control variables in model (2). The control variables are gender, age, residence in an urban or small town (vs. the omitted category of large town), country, marital status, occupation, age of the respondent at which her full-time education ended (as an indicator for education), and fixed telephone line and mobile telephone ownership as a proxy for wealth. The same models are also run for the subsample of individuals who are willing to donate or not willing to donate. We also experiment with different samples. We exclude Spain from the sample to see whether this produces any changes. All regression models are estimated with standard errors clustered at the country level. For ease of interpreting the coefficients, marginal effects calculated from the probit estimation results are displayed in Table IV. We observe significantly lower organ donation card holding in presumed consent countries on average. In line with our expectations, we find lower donation card holding among individuals who are willing to donate their organs in presumed consent countries compared with informed consent countries. However, surprisingly, among people who are not willing to donate, we do not observe higher registration in presumed consent countries. These findings imply that presumed consent legislation is likely to increase cadaveric donation rates because unwilling individuals fail to register their preferences in presumed consent countries. There could be many reasons for this. For instance, informed consent countries might be more active in soliciting organ donation through advertising the donation card more actively. Another channel could be different registration costs in different countries.15 For the purposes of this study, it suffices to observe higher organ donation card holding for both individuals who are willing to donate and not willing to donate in informed consent countries. 14
For example, if a person did not register on a non-willing list in an informed consent country, family members might be more likely to give consent, because this signals that the person is not strongly against organ donation. 15 For instance, in the Netherlands and in the UK, both informed consent countries, online registration of preferences for organ donation is possible. But Sweden, a presumed consent country, also allows for online registration of preferences. Copyright © 2014 John Wiley & Sons, Ltd.
Health Econ. 24: 1560–1572 (2015) DOI: 10.1002/hec
1565
PRESUMED CONSENT
Table IV. Organ donation card holding regression results Total sample
a. All countries Presumed Baseline controls N b. Excluding Spain Presumed Baseline controls N
Willing to donate
Not willing to donate
(1)
(2)
(1)
(2)
(1)
(2)
0.130** (0.057) – 27,584
0.118** (0.047) + 27,167
0.224*** (0.070) – 15,053
0.214*** (0.062) + 14,880
0.015 (0.019) – 7793
0.012 (0.014) + 7643
0.130** (0.057) – 26,558
0.119** (0.047) + 26,167
0.224*** (0.071) – 14,487
0.215*** (0.063) + 14,333
0.015 (0.019) – 7610
0.012 (0.014) + 7464
Standard errors (in parentheses) are clustered at country level, *p < .1, **p < .05, ***p < .01. Baseline controls: gender, age, age, residence in an urban or small town (vs. the omitted category of large town), country, marital status, occupation, age of the respondent at which her full-time education ended, and fixed telephone line and mobile telephone ownership.
2.3. The impact of presumed consent on cadaveric donors and transplantations 2.3.1. Descriptive statistics. Figures 1 and 2 show higher cadaveric donor and kidney transplantation rates in presumed consent countries in comparison with informed consent countries. Yet, Bulgaria, having the lowest levels of cadaveric donors and kidney transplantations, is a presumed consent country. This suggests that there are other factors that need to be taken into account. Mainly following previous literature, information on a number of factors, such as number of deaths by specific causes, health spending, medical infrastructure, religious beliefs and education level, is gathered from a variety of sources.16 In most cases, deceased donors were brain-dead and their hearts were artificially functioning with the help of ventilation machines. The most common causes of brain death are homicides, motor vehicle accidents and cerebro-vascular diseases. According to the Eurobarometer 2009 survey, religious reasons, distrust in the system and fear of manipulation of the human body are three major causes of refusal to donate organs. To capture trust in the system, corruption perception scores from Transparency International are included. Corruption perception scores range between 10 (highly clean) and 0 (highly corrupt). To control for religiosity changes over time, we compiled percentage of population being Roman Catholic and having no religion17 from four surveys: the International Social Survey Program, the European Social Survey, the European Values Survey and Eurobarometers conducted between 1999/2000 and 2010. Table V provides summary statistics for the sample. The second column shows presumed consent countries having 16.95 cadaveric donors pmp per year, whereas the third column shows informed consent countries having 13.17 cadaveric donors pmp per year. From the fourth column, we observe presumed consent countries have 3.8 more cadaveric donors pmp per year, and this difference is statistically significant. We also observe significantly higher cadaveric donation after brain death and kidney transplantation rates in presumed consent countries. There are other differences between these countries. In particular, deaths from homicide, motor vehicle accidents and cerebro-vascular diseases are higher in presumed consent countries. However, only the motor vehicle accident difference is statistically significant. Health expenditure per capita is relatively lower, whereas the number of hospital beds per capita is relatively higher in presumed consent countries. Informed consent countries, having higher corruption perception scores, are perceived as more transparent. No major difference is observed in the percentage of the population considering themselves as having no religion, while we observe 16 17
Detailed information on sources of the data is presented in Appendix D. Includes people who consider themselves as agnostic, atheist or having no religion.
Copyright © 2014 John Wiley & Sons, Ltd.
Health Econ. 24: 1560–1572 (2015) DOI: 10.1002/hec
1566
Z. B. UGUR
Figure 1. Cadaveric organ donor rate per million population in 2010
Figure 2. Kidney transplantation from deceased donors rate per million population in 2010
a higher percentage of the population considering themselves as Roman Catholic in presumed consent countries. Relative to informed consent countries, presumed consent countries have a statistically lower percentage of upper secondary or tertiary educated male population. No significant difference was found for the upper secondary or tertiary educational attainment of females. 2.3.2. Regression output for cadaveric donation rate. In this section, we provide a regression analysis for total cadaveric donation and brain death donation rate. We differentiate between donation after brain death (known as heart-beating donation) and donation after circulatory death (known as non-heart-beating donation) because the latter is often less efficient for transplantation (Cota et al., 2013). Table VI shows the regression output for the log of total cadaveric donor rate. Model (1) contains only a dummy variable for presumed consent legislation. In model (2), to address potential confounding factors, we Copyright © 2014 John Wiley & Sons, Ltd.
Health Econ. 24: 1560–1572 (2015) DOI: 10.1002/hec
1567
PRESUMED CONSENT
Table V. Descriptive statistics (means and standard deviations for 2000–2010)
Presumed consent country Cadaveric donor rate, pmp Cadaveric donation after brain death, pmp Kidney transplant rate, pmp Homicide, pmp Deaths from motor vehicle accident rate, pmp Deaths from cerebro-vascular diseases, pmp Health expenditure per capita Hospital beds, per 100,000 population Corruption perception score % of people having no religion % of people being Roman Catholic % of higher educational attainment, males % of higher educational attainment, females Number of countries
Entire sample
Presumed consent
Informed consent
Difference
0.74 [0.44] 15.96 [7.65] 15.40 [7.36] 25.03 [11.38] 21.48 [25.80] 113.32 [48.99] 853.95 [498.73] 2238.78 [1260.52] 580.83 [166.77] 6.28 [1.99] 0.23 [0.18] 0.42 [0.35] 68.62 [13.13] 67.49 [12.07] 27
16.95 [7.96] 16.58 [7.65] 26.54 [11.75] 21.64 [25.96] 119.08 [42.60] 867.44 [473.35] 2144.57 [1236.08] 582.62 [160.31] 5.98 [1.84] 0.23 [0.18] 0.45 [0.35] 67.45 [14.62] 66.70 [13.38] 20
13.17 [5.88] 12.09 [5.24] 20.68 [8.94] 21.01 [25.48] 97.47 [60.91] 816.88 [564.44] 2507.97 [1298.67] 575.72 [185.03] 7.12 [2.16] 0.22 [0.17] 0.34 [0.34] 71.95 [6.39] 69.73 [6.74] 7
3.78*** (1.00) 4.49*** (0.95) 5.85*** (1.48) 0.63 (3.45) 21.61*** (6.45) 50.55 (66.86) 363.40* (165.84) 6.90 (22.26) 1.14*** (0.26) 0.01 (0.02) 0.12* (0.05) 4.50** (1.72) 3.03 (1.59)
Notes: standard deviations in [ ]; standard errors in ( ). *p < 0.05, **p < 0.01, ***p < 0.001.
Table VI. Pooled ordinary least squares estimates of log cadaveric donor rate (1) Legislation Presumed consent Practicing legislation Family consent Presumed consent * family consent Potential donors Log of MVA + CVD + homicide, pmp Health spending Log of health expenditure per capita Medical infrastructure Log of hospital beds per 100,000 people Trust in the system Corruption perception score Religious beliefs % of people having no religion % of people being Roman Catholic Education % of higher educational attainment, males % of higher educational attainment, females Average willingness to donate Country group fixed effects R-squared N
0.355 (0.401)
(2) 0.329** (0.157)
(3) 0.390** (0.148)
(4) 0.350* (0.190)
(5) 0.042 (0.163) 0.309 (0.341) 0.552* (0.291)
– 0.037 287
0.038 (0.076)
0.072 (0.076)
0.108 (0.072)
0.072 (0.068)
0.227 (0.181)
0.212 (0.185)
0.190 (0.212)
0.190 (0.162)
0.073 (0.273)
0.226 (0.258)
0.092 (0.280)
0.151 (0.241)
0.117*** (0.042)
0.160*** (0.048)
0.079 (0.059)
0.071 (0.054)
1.824*** (0.345) 1.237*** (0.254)
1.818*** (0.347) 1.278*** (0.255)
1.914*** (0.508) 1.353*** (0.355)
1.873*** (0.487) 1.381*** (0.306)
0.038** (0.017)
0.040** (0.015)
0.007 (0.020)
0.000 (0.019)
0.043** (0.018)
0.045** (0.017)
0.007 (0.025)
0.001 (0.022)
– 0.655 262
1.153 (0.809) – 0.664 262
+ 0.741 262
+ 0.755 262
Standard errors (in parentheses) are clustered at country level, *p < .1, **p < .05, ***p < .01.
include control variables such as death rates from homicides, motor vehicle accidents and cerebro-vascular diseases, health expenditure per capita, hospital beds per 100,000 population, corruption perception scores, percentage of population considering themselves as having no religion and Roman Catholic, and Copyright © 2014 John Wiley & Sons, Ltd.
Health Econ. 24: 1560–1572 (2015) DOI: 10.1002/hec
1568
Z. B. UGUR
percentage of upper secondary or tertiary educated males and females. According to model (2), the coefficient of the presumed consent indicates a 33% higher cadaveric donor rate. This difference is significant at the 5% level. Model (3) includes average willingness to donate organs in each country from the Eurobarometer surveys conducted in 2002, 2006 and 2009, to address the potential confounding effect of social acceptance of organ donation. The coefficient of presumed consent dummy is still significant, suggesting that the impact of presumed consent on the cadaveric donor rate cannot be attributed to higher social acceptance of organ donation. To capture the idea that presumed consent might be enacted in countries where there is different unobserved heterogeneity, we group countries in model (4) according to the following classification, with presumed consent countries in bold: (1) (2) (3) (4) (5) (6) (7) (8) (9)
Ireland–UK Greece–Cyprus Germany–Austria–Hungary (Hungary is included here because of the historical connection between Hungary and Austria) Netherlands–Belgium–Luxembourg Estonia–Lithuania–Latvia (These countries together formed the Balttransplant organization) Denmark–Sweden–Finland (which formed the Scandiatransplant organization) Poland–Czech Republic–Slovakia (Western Slavic ethnic origin) Portugal–Spain–Italy–France (Latin ethnic origin) Bulgaria–Romania–Croatia–Slovenia (Southern-Slavic ethnic origin. These countries are also known as Balkan countries)
Model (4) shows 35% higher donation rates in presumed consent countries when country group dummy variables are included in the model. In model (5), we included family consent and its interaction with presumed consent to check whether family consent makes any difference. Model (5) suggests that presumed consent does matter especially when combined with family consent, whereas in informed consent countries, seeking family consent has a negative but insignificant coefficient. This result is in line with the prediction that even if families make the final decision on organ donation, presumed consent can lead to notably higher donation rates. The same five models of the cadaveric donor rate are also estimated for the log of brain death donor rate. Table VII provides the regression output. Overall, the regression results suggest a more pronounced impact of presumed consent.
2.3.3. Regression output for kidney transplantation rate. In this section, we provide a regression analysis for the kidney transplantation rate. We use the same five models of the cadaveric donor rate to analyze the log kidney transplant rate. Table VIII provides the regression results. Model (1) shows that presumed consent countries have a 33% higher kidney transplant rate. Similar to cadaveric donor regression results, models (2) and (3) report that presumed consent countries have statistically significantly higher kidney transplant rates. In model (4), when country group dummy variables are included, presumed consent countries are found to have a 31% higher kidney transplant rate. A similar pattern emerges from model (5) as with the cadaveric donor rate. Presumed consent increases the kidney transplant rate when combined with family consent. In line with intuition, family consent in informed countries has a negative yet insignificant coefficient. Overall, the regression results from Tables VI and VIII suggest that presumed consent countries have 28% to 32% higher cadaveric donation and 27% to 31% higher kidney transplant rates than informed consent countries. We conducted a number of robustness checks, such as inclusion of year fixed effects and exclusion of Spain from the sample. The weighted least squares regression is also estimated, in which weighting proportional to size of population is used. Details of these tests are reported in Appendix E. Copyright © 2014 John Wiley & Sons, Ltd.
Health Econ. 24: 1560–1572 (2015) DOI: 10.1002/hec
1569
PRESUMED CONSENT
Table VII. Pooled ordinary least squares estimates of log of brain death donor rate
Legislation Presumed consent Practicing legislation Family consent * Presumed consent family consent Baseline controls Country group fixed effects R-squared N
(1)
(2)
(3)
(4)
(5)
0.408 (0.388)
0.386** (0.151)
0.422*** (0.144)
0.366** (0.161)
0.005 (0.136) 0.372 (0.279) 0.647** (0.245)
– – 0.050 287
+ – 0.665 262
+ – 0.668 262
+ + 0.762 262
+ + 0.781 262
Standard errors (in parentheses) are clustered at country level, *p < .1, **p < .05, ***p < .01. Baseline controls: death rates from homicides, motor vehicle accidents and cerebro-vascular diseases, health expenditure per capita, hospital beds per 100,000 population, corruption perception scores, percentage of population considering themselves as having no religion and Roman Catholic, and percentage of upper secondary or tertiary educated males and females.
Table VIII. Pooled ordinary least squares estimates of log kidney transplant rates
Legislation Presumed consent Practicing legislation Family consent * Presumed consent Family consent Baseline controls Country group fixed effects R-squared N
(1)
(2)
(3)
(4)
(5)
0.328 (0.387)
0.347** (0.154)
0.367** (0.153)
0.309* (0.178)
0.015 (0.141) 0.180 (0.315) 0.525* (0.278)
– – 0.030 296
+ – 0.614 270
+ – 0.615 270
+ + 0.737 270
+ + 0.756 270
Standard errors (in parentheses) are clustered at country level, *p < .1, **p < .05, ***p < .01. Baseline controls: death rates from homicides, motor vehicle accidents and cerebro-vascular diseases, health expenditure per capita, hospital beds per 100,000 population, corruption perception scores, percentage of population considering themselves as having no religion and Roman Catholic, and percentage of upper secondary or tertiary educated males and females.
3. DISCUSSION In this section, firstly, we discuss further whether presumed consent is an indicator of a country’s commitment to organ donation rather than a causal mechanism in itself. Secondly, we discuss the advantages and disadvantages of both types of consent legislation. A country’s commitment to organ donation could be indicated by available transplant capacity. In our main analysis, transplantation capacity is imperfectly proxied by the number of hospital beds per 100,000 people. Models (1) and (2) in Table IX use the same specifications as models (3) and (4) in Table VI, with the logarithm of number of transplantation centers pmp as an additional control variable. Table IX reports an increased coefficient of presumed consent compared with Table VI. Kidney transplant capacity (defined as the number of kidney transplant centers pmp) could also influence kidney transplantation rates. To check this idea, models (3) and (4) in Table IX use the same specifications as models (3) and (4) in Table VIII, with the logarithm of number of transplantation centers pmp as an additional control variable. Models (3) and (4) also report an increased coefficient of presumed consent compared with Table VIII. Copyright © 2014 John Wiley & Sons, Ltd.
Health Econ. 24: 1560–1572 (2015) DOI: 10.1002/hec
1570
Z. B. UGUR
Table IX. Transplant infrastructure and presumed consent (pooled ordinary least squares) Log cadaveric donor rate
Legislation Presumed consent Transplant Infrastructure Log of kidney transplant centers Baseline controls Average willingness to donate Country group fixed effects R-squared N
Log kidney transplant rate
(1)
(2)
(3)
(4)
0.546*** (0.147)
0.683*** (0.164)
0.494*** (0.139)
0.561*** (0.161)
0.299** (0.128) + + – 0.778 153
0.274 (0.172) + – + 0.828 153
0.352** (0.141) + + – 0.772 153
0.322** (0.153) + – + 0.836 153
Standard errors (in parentheses) are clustered at country level, *p < .1, **p < .05, ***p < .01. Baseline controls: death rates from homicides, motor vehicle accidents and cerebro-vascular diseases, health expenditure per capita, hospital beds per 100,000 population, corruption perception scores, percentage of population considering themselves as having no religion and Roman Catholic, and percentage of upper secondary or tertiary educated males and females.
On the whole, the results of Table IX suggest that the regression results provided in Tables VI and VIII are not confounded by commitment to organ donation differences across countries. As regards disadvantages, both presumed consent and informed consent legislation are error-prone. In presumed consent legislation, non-willing individuals’ organs could be mistakenly removed if they did not register their non-willingness (Gill, 2004; Orentlicher, 2008). In this case, removal of an organ is wrong because it is against the patient’s wishes. In informed consent countries, willing individuals’ organs could be mistakenly not removed if they did not register their willingness (Gill, 2004; Orentlicher, 2008). In this case, non-removal of an organ is a waste of a scarce resource that could be used for improving the life of another person. So, in both scenarios, there is a possibility that some people’s wishes may not be respected. Using descriptive information from Tables I and III, we can calculate which regime produces the least percentage of errors and which regime maximizes the percentage of people whose wishes are respected. The probability of making a mistake in presumed consent legislation is 43%, which is calculated by multiplying the percentage of non-willing individuals with the probability of non-registering their preference (= (100 55.83) * (100 1.56) / 100). The probability of making a mistake in informed consent legislation is 37.71, which is given by multiplying the percentage of willing individuals with the probability of nonregistering their preference (= 55.56 * (100 32.11) / 100). If a default consent regime should be chosen such that the least number of errors is made, an informed consent regime seems to be superior. The percentage of people whose wishes are respected in an informed consent regime is those who are not willing to donate (25.45%) plus those who are willing to donate and registered their preference (= 55.56 * 32.11 / 100). This corresponds to 43.29%. The percentage of people whose wishes are respected in a presumed consent regime is those who are willing to donate (55.83%) plus those who are not willing to donate and registered their preference (= 26.99 * 1.56 / 100). This corresponds to 56.25%. If the criterion for choosing the default consent regime is to maximize the percentage of people whose wishes are respected, presumed consent seems to outperform the informed consent regime. As regards advantages, presumed consent may easily outperform informed consent legislation in terms of administration costs. Although we do not have data about how much it costs to administer presumed consent and informed consent legislation, it is highly probable that an informed consent regime is more costly because the organ donation authority of an informed consent country has to invest sizeable amounts of money in convincing people to be organ donors. Moreover, these investments have a very low probability of return, because the likelihood of dying under conditions which would be suitable for organs to be transplanted is very low. Howard and Byrne (2007) estimated the probability of a potential donor being an actual donor at some point in her lifetime to be 0.0028. Copyright © 2014 John Wiley & Sons, Ltd.
Health Econ. 24: 1560–1572 (2015) DOI: 10.1002/hec
PRESUMED CONSENT
1571
In brief, if policy makers attribute greater importance to minimizing the number of errors, informed consent seems to be preferable. If policy makers attribute greater importance to saving lives, respecting people’s wishes or minimizing administration costs, presumed consent seems to be preferable.
4. LIMITATIONS Currently, the Eurobarometer surveys are the only source that allows cross-country comparisons on organ donation preferences. However, they do have some limitations. Firstly, the organ donation preferences are based on self-reporting by survey respondents. The quality of the information may therefore be affected by over-reporting due to social desirability. Our analysis in this study is based on the assumption that people in presumed and informed consent countries over-report their willingness to donate by the same amount. However, preference for organ donation in a country is potentially affected by the consent law because this sends a signal about the social norm on organ donation. Therefore, measures of willingness to donate will appear higher for people from presumed consent countries than informed consent countries. In this case, our results on which consent regime produces the least percentage of errors might be misleading. Secondly, the question about organ donation preference is asked in terms of whether the respondent wants to donate one of her organs or not. A person may very well agree to donate some of her organs but not agree to donate other organs. Unfortunately, the dataset does not allow tracking of organ donation preferences for specific organs. In reality, organ donation cards often let people declare which organs they want to donate or not. However, the organ donation card question is formulated as a 0/1 type question, making it impossible to comment on partial donation registration. Thirdly, the wording of the organ donation card question might lead respondents to think that having the card only refers to people who have declared that they want to be donors. A better wording could be, ‘Do you have an organ donation declaration or refusal card/ registration?’
5. CONCLUSION In this study, we attempt to extend the literature on how presumed consent impacts cadaveric donors and kidney transplantations in the EU-27 countries plus Croatia in the period 2000–2010. As a first step, we show evidence that presumed consent is not necessarily legislated in countries where there is higher social acceptance of organ donation. In the second step, we show that people fail to take action in line with their preferences, which is especially relevant for individuals who are unwilling to donate their organs. Therefore, we suggest that presumed consent is likely to produce more organ donors. In the main analysis, after accounting for potential confounding factors, our estimates suggest that presumed consent countries have 28% to 32% higher cadaveric donation and 27% to 31% higher kidney transplant rates than informed consent countries. Our study indicates that presumed consent legislation can be instrumental in saving lives, respecting people’s wishes and minimizing the administration costs.
CONFLICT OF INTEREST The authors have no conflict of interest.
ACKNOWLEDGEMENTS
I thank Peter Kooreman, Shelly Lundberg, Jan van Ours, Marcel Das, Matthijs Kalmijn, seminar participants at Tilburg University and Lowland Health Economists’ Study Group (LolaHESG). Copyright © 2014 John Wiley & Sons, Ltd.
Health Econ. 24: 1560–1572 (2015) DOI: 10.1002/hec
1572
Z. B. UGUR
REFERENCES
Abadie A, Gay S. 2005. The impact of consent legislation on cadeveric organ donation: a cross country study. NBER Working Paper No. 10604. Becker G, Elías J. 2007. Introducing incentives in the market for live and cadaveric organ donations. The Journal of Economic Perspectives 21: 3–24. Bilgel F. 2010. The impact of presumed consent laws and institutions on deceased organ donation. The European Journal of Health Economics 13: 29–38. Chang G, Mahanty H, Ascher N, Roberts J. 2003. Expanding the donor pool: can the Spanish model work in the United States? American Journal of Transplantation 3: 1259–1263. Cohen LR. 1989. Increasing the supply of transplant organs: the virtues of a futures market. George Washington Law Review 58: 1–51. Cota N, Burgess M, English W. 2013. Organ donation after circulatory death. Anaesthesia Tutorial of the Week 282: 2. Davis CL, Delmonico FL. 2005. Living-donor kidney transplantation: a review of the current practices for the live donor. Journal of American Society of Nephrology 16: 2098–2110. Ellison M, McBride M, Taranto S, Delmonico F. 2002. Living kidney donors in need of kidney transplants: a report from the Organ Procurement and Transplantation Network. Transplantation 74: 1349–1351. Gevers S, Janssen A, Friele R. 2004. Consent systems for post mortem organ donation in Europe. European Journal of Health Law 11: 175–186. Gill M. 2004. Presumed consent, autonomy, and organ donation. Journal of Medicine and Philosophy 29: 37–59. Global Observatory on Donation and Transplantation. 2011, February 9. Retrieved from http://data.transplant-observatory. org/paginas/informes/DatosUsuario.aspx Gnant M, Wamser P, Goetzinger P, Sautner T, Steininger R, Muehlbacher F. 1991. The impact of the presumed consent law and a decentralized organ procurement system on organ donation: quadruplication in the number of organ donors. Transplant Proceedings 23: 2685–2686. Greene W. 2011. Fixed Effects vector decomposition: a magical solution to the problem of time invariant variables in fixed effects models? Political Analysis 19: 135–146. Healy K. 2005. The political economy of presumed consent. Working Paper 31, University of California, Los Angeles. Howard D. 2007. Producing organ donors. The Journal of Economic Perspectives 21: 25–36. Howard DH, Byrne MM. 2007. Should we promote organ donor registries when so few registrants will end up being donors? Medical Decision Making 27: 243–249. Johnson E, Goldstein D. 2003. Do defaults save lives? Science 302: 1338–1339. Matesanz R, Miranda B. 2002. A decade of continuous improvement in cadaveric organ donation: the Spanish model. Journal of Nephrology 15: 22–28. Newsletter Transplant. 2009. http://www.ont.es/publicaciones/Paginas/Publicaciones.aspx. (accessed February 11, 2011) Newsletter Transplant. 2010. http://www.ont.es/publicaciones/Paginas/Publicaciones.aspx. (accessed February 11, 2011) Orentlicher D. 2008. Presumed consent to organ donation: its rise and fall in the United States. Rutgers Law Review 61: 295–331. Oz M, Kherani A, Rowe A, Roels A, Crandall C, Tomatis L, et al. 2003. How to improve organ donation: results of the ISHLT/FACT poll. Journal of Heart and Lung Transplantation 22: 389–410. Paparde I. 2010. Latvian public unwilling to donate organs. Retrieved 9 29, 2012, from http://ec.europa.eu/health/ blood_tissues_organs/docs/art_latvia_2010_en.pdf Roels L, Vanrenterghem Y, Waer M, Christiaens M, Gruwez J, Michielsen P. 1991. 3 years of experience with a presumed consent legislation in Belgium—its impact on multiorgan donation in comparison with other European countries. Transplantation Proceedings 23: 903–904. Roth A. 2007. Repugnance as a constraint on markets repugnance as a constraint on markets. The Journal of Economic Perspectives 21: 37–58.
SUPPORTING INFORMATION Additional supporting information may be found in the online version of this article at the publisher's web site.
Copyright © 2014 John Wiley & Sons, Ltd.
Health Econ. 24: 1560–1572 (2015) DOI: 10.1002/hec
