BSc Thesis Economics and Business
Economics
The impact of presumed consent legislation on family refusal
rates: Towards a new organ donor registration system in the
Netherlands
Written by: Viktor Kelderman Student Number: 10643400 Supervisor: Andro Rilović
2
Table of Contents
Abstract 3 I. Introduction 4 II. Methodology 7 III. Results 11 IV. Discussion 15A. The Dutch Case 16
V. Concluding Remarks 20
Appendix 21
References 24
Statement of Originality
This document is written by Viktor Kelderman who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
3 Abstract
A lot of research has focused on the influence of presumed consent legislation on deceased donation rates. Family consent, which is almost always needed before donation can take place, has due to a lack of data not been properly accounted for. In this paper the impact of consent legislation on family refusal rates is estimated. Using pooled OLS and a sample of 32 countries during the period 2003-2015, the author finds that presumed consent has a negative and sizeable effect on family refusal rates. However, after controlling for country group fixed effects the impact of legislation becomes negligible. To address additional factors with influence on refusal rates that are only at play when a switch in the consent system is enacted, Chile and Venezuela are examined. It is concluded that loss aversion and mistrust towards the health care system may lead to an increase in refusal rates when a country moves from informed- to presumed consent. With these results in hand a simple way to determine the impact of a switch in consent legislation on deceased donor rates is proposed that is tested on the Netherlands. An educated guess reveals that the proposed system will at best have a slight positive effect on deceased donor rates.
4 I. Introduction
During the period of 2010-2015 roughly 850 persons on the organ waiting list in the Netherlands lost their lives (Newsletter Transplant, n.d.). In an effort to alleviate the shortage of donors the Lower House recently voted1 in favour of a proposal that aims at changing the present deceased donor registration system, a system of informed consent, to a presumed consent system wherein a person who fails to make a choice is registered as having ‘no objection’ to donating (Dool & Steenbergen, 2016). This alteration should increase donation rates2 because of the different properties of the present and proposed registration system. In an informed consent system a donor needs to explicitly authorise the removal of any organs after his death for donation, usually through the recording of consent in a national registry. In a presumed system no explicit consent is required. It is sufficient that a person did not object during life to becoming a donor after death (Coppen, Friele, Gevers, & Marquet, 2005). In other words, by default a person is a non-donor under an informed consent system, but a donor under a presumed consent system. This difference in the default rule is crucial as it plays an important part in the decision-making of individuals (Sunstein & Thaler, 2003). People seem to stick to the default for a variety of reasons, which range from simple inertia caused by procrastination and sloth to loss aversion (Kahneman, Knetsch, & Thaler, 1991; Sunstein & Thaler, 2003, p. 1181). In the case of organ donation the status-quo bias might be exacerbated, because a decision on organ donation is connected to one’s own death and as such might pose significant mental costs (Ugur, 2015, pp. 1560-1561).
The effect of the default rule is visible in the Netherlands. Almost 60% of eligible citizens have not registered a choice on organ donation (NTS, n.d.). If a large part of this group were to stick to the new default then donation rates would be expected to increase and waiting lists to decrease quite drastically. There is literature that suggests that presumed consent legislation indeed has a positive effect on donation rates (Goldstein & Johnson, 2004; Abadie & Gay, 2006; Ugur, 2015). However, opposing research exists that does not share this result (Coppen, et al., 2005). A possible reason for not observing any differences in donation rates could be connected to family consent. The next of kin’s decision to agree to donation might be influenced by the degree of explicitness of the consent to donate by the deceased.
1 Before the implementation of the bill can officially take place a majority vote is needed in the Upper House as
well. The Upper House can only accept or reject the bill, it does not have the authority to make amendments.
5 Family members seem to be more willing to give consent when someone is an explicit donor than when a person is a presumed donor (Coppen, Friele, Gevers, & van der Zee, 2010; Faun, et al., 2014). However, Ugur (2015) suggests that presumed consent legislation might have a positive effect on family consent, because in explaining the need for donation to the family of the deceased doctors have the law on their side (p. 1561). She does not clarify what exactly is meant by ‘doctors have the law on their side’. It would make the most sense if we treat this point as a twist on the previous argument. Under presumed consent all people are registered. Even if the decision to donate is not explicit then family might still sooner give consent for a presumed donor than for someone who is not registered at all. Consequently, if the group of unregistered people under an informed consent system is large enough then we would observe an overall higher family refusal rate than under presumed consent3.
Both arguments are dependent on the role and the influence on the donation decision that the family has. This role is also paramount to the theoretical argument in which a presumed consent system is thought to increase donation rates. If the decision on donation is a decision made by the family of the deceased as much as by the deceased himself then changing the default rule might have a far less pronounced effect. In reality the next of kin are almost always consulted, regardless of the system in place, and have a final say in the organ procurement process (Boyarsky, et al., 2012; Beitel, et al., 2012). Ass family consent is pivotal it stands to reason to further assess the validity of both claims.
In this paper several contributions will be made to existing literature. First and foremost, the effect of consent legislation on family refusal rates will be estimated using a panel of countries for which data on family consent are available. Previous research has made use of results gathered through surveys (Coppen, et al., 2010; Ugur, 2015). A donation decision is a very stressful and difficult4 process connected to the loss of someone close. There is little reason to think that the outcome in reality can be adequately approximated for in a controlled environment. Conclusions drawn from these results therefore lack validity.
A cross-country study runs the risk of overlooking dynamics in refusal rates that are only at play when the consent system is actually changed. To see if such dynamics are present
3 This argument is found in formula form in the Discussion
6 Chile and Venezuela are examined before and after they changed from informed to presumed consent.
Because the found results are likely to carry an implication for consent policy some space is alotted to see what results can be expected from the proposal in the Netherlands that aims at changing donor legislation. As is shown in Figure 1, family refusal rates in the Netherlands have been very high over the past 10 years and have led to the loss of countless potential donor organs. In light hereof the kind of effect that the switch to presumed consent has on family refusal rates is particularly relevant for the Netherlands.
Figure 1. Family Refusal Rates and Unused Potential Donors in the Netherlands from 2006-20155
5 The increase in unused potential donors is the result of an increase in the number of potential donors.
𝑈𝑛𝑢𝑠𝑒𝑑 𝑃𝑜𝑡𝑒𝑛𝑡𝑖𝑎𝑙 𝐷𝑜𝑛𝑜𝑟𝑠 = 𝑃𝑜𝑡𝑒𝑛𝑡𝑖𝑎𝑙 𝐷𝑜𝑛𝑜𝑟𝑠 × 𝐹𝑎𝑚𝑖𝑙𝑦 𝑅𝑒𝑓𝑢𝑠𝑎𝑙 𝑅𝑎𝑡𝑒 0,00% 10,00% 20,00% 30,00% 40,00% 50,00% 60,00% 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 0 50 100 150 200 250 300 350 400 450 Fami ly Re fu sal Rat e (Lin e) U n u se d Po ten tia l Do n o rs (Bars)
7 II. Methodology
To estimate the effect of consent legislation on family refusal rates a panel dataset of countries was constructed. The Newsletter Transplant, which publishes a detailed yearly report on worldwide transplant activities, was used as the main source of information on refusal rates per country. The publications accessible online span a 13-year period (2003-2015). Therefore, the same period was analysed in this paper. For the Netherlands, Austria, Germany, and the USA, these numbers were supplemented with the help of national reports6 on transplant activities. A sample of 58 countries with data on family refusal rates was the result. However, 26 were dropped. For what reasons will become clear in a moment. The 32 countries in the final sample were categorized according to the type of consent system in place, as can be seen in Table 1 in the Appendix.
The countries left out from the analyses were left out because they either had: a mixed consent system, less than three data points7, switched consent system in the 13 year period, had too little observations per year, or had inconsistent consent legislation. These criteria, notwithstanding their respective differences, were set to keep estimation results from being biased. A listing of which countries were excluded and for what reason is given in Table 2 in the Appendix.
The final sample was unbalanced. The mean number of data points on family refusal rates
per country was 7.9. If these data were missing because of a selection process related to the value of the dependent variable, beyond depending on the explanatory variable, then sample selection bias could be introduced to the estimator (Stock & Watson, 2015, p. 372). Based on the composition of the sample there is little reason to think that either a presumed – or an informed consent country is more likely to report on family refusal rates. The ratio of countries with a presumed consent system to those with an informed one (≈3:2) in the sample is consistent with previous literature (Abadie & Gay, 2006).
6 For the Netherlands, Austria, Germany, and the USA, these national reports are: NTS Jaarverslag, Transplant
Jahresbericht, Organspende und Transplantation in Deutschland Jahresbericht, and The Annual Data Report by the American Journal of Transplantation.
7 Longitudinal models often require at least three repeated measures (Ferguson, O'Carroll, & Shepherd, 2014,
8 The model used for the analyses was, following the research of Ugur (2015) and Abadie & Gay (2006), a pooled OLS estimator. Instead of the log of donation rates the log8 of the family refusal rate functioned as the dependent variable. The consent system was kept as the main explanatory variable.
(1) ln (𝐹𝑎𝑚𝑖𝑙𝑦 𝑅𝑒𝑓𝑢𝑠𝑎𝑙 𝑅𝑎𝑡𝑒𝑖𝑡) = 𝛽0+ 𝛽1𝐶𝑜𝑛𝑠𝑒𝑛𝑡𝑖𝑡+ 𝑢𝑖𝑡 𝑤ℎ𝑒𝑟𝑒, 𝑖 = 𝐴𝑟𝑔𝑒𝑛𝑡𝑖𝑛𝑎, … , 𝑈𝑆𝐴; 𝑡 = 2003, … , 2015; 𝑎𝑛𝑑
𝐶𝑜𝑛𝑠𝑒𝑛𝑡𝑖𝑡 = {
1 𝑖𝑓 𝑐𝑜𝑢𝑛𝑡𝑟𝑦 𝑖 ℎ𝑎𝑠 𝑎 𝑝𝑟𝑒𝑠𝑢𝑚𝑒𝑑 𝑐𝑜𝑛𝑠𝑒𝑛𝑡 𝑠𝑦𝑠𝑡𝑒𝑚 𝑖𝑛 𝑦𝑒𝑎𝑟 𝑡 0 𝑖𝑓 𝑐𝑜𝑢𝑛𝑡𝑟𝑦 𝑖 ℎ𝑎𝑠 𝑎𝑛 𝑖𝑛𝑓𝑜𝑟𝑚𝑒𝑑 𝑐𝑜𝑛𝑠𝑒𝑛𝑡 𝑠𝑦𝑠𝑡𝑒𝑚 𝑖𝑛 𝑦𝑒𝑎𝑟 𝑡
Preferable to pooled OLS would have been a fixed effects model. However, a fixed effects model requires a group of countries that have switched consent systems in years for which data are available. As is visible in Table 2, only Chile and Venezuela switched systems during the observed period. This is clearly an insufficient sample size. A fixed effects model was impractical for the current sample because it causes all time invariant variables to drop out and therefore also our main variable of interest.
Before OLS could be proceeded with two potential problems needed to be addressed: heteroscedasticity and autocorrelation. In order to test for autocorrelation a method by Woolridge (2002, pp. 282-283) suitable for panel-data models was used. The produced test-statistic was insignificant, indicating the absence of autocorrelation. A plot of the residuals and their fitted values on the other hand did reveal signs of heteroscedasticity, as did formal tests like the Breusch-Pagan/Cook-Weisberg and White’s test. Robust standard errors9 solve the issue, but in the case of heteroscedasticity OLS is no longer BLUE. A different model might therefore have been more efficient. Determining what this model should have looked like based on the nature of the heteroscedasticity was outside the scope of this research. Still, the inclusion of robust standard errors should have solved for bias in the errors and subsequent bias in the significance tests of the estimated coefficients.
8 The log was taken to get the effect of consent legislation on refusal rates in percentages.
9 The simple model in equation (1) was expanded upon by including additional variables to account for two often heard reasons for refusing organ donation: religious beliefs and distrust in the system (European Commision, 2010).
For religious beliefs data on the percentages of Roman Catholics, Muslims10, and non-religious people of a country’s population were used. These were the only groups that were consistently reported on for all countries. For the retrieval of the data the Central Intelligence
Agency World Factbook and the summaries by the U.S. Department of State were consulted.
The latter was the leading source when information between the two differed. All percentages were based on demographical estimates in some year between 2003-2015. The year decided upon equalled the year for which an estimate of a country’s religious population was available in the database. Because often only one estimate per country was available the percentages were assumed to be fixed during the rest of the years11.
To measure distrust in the system the scores on the Corruption Perceptions Index ,published yearly by Transparency International, were used. These scores are measured on a scale of 1 to 10. A 10 represents a very ‘clean’ country and a 1 a country that is perceived as extremely corrupt. This measure is a fairly broad one, especially when researching the consent system and not the system in general. There is a simple reason for its use. In the Eurobarometer survey, respondents were asked not about distrust in the consent - or even the health care system, but about distrust in the system in general. As stated in the report ‘’this could include the transplantation system, consent system or in general the society system’’ (European Commision, 2010, p. 28). Therefore, a more general measure of distrust is appropriate.
10 Whereas the percentage of Roman Catholics is standardly used when estimating the influence of religion on
organ donation, the percentage of Muslims is not. The Muslim population was included in the regression because Islam is not entirely clear on the justifiability of organ donation, in particular deceased donation (Ahmed, Oliver, Saif, & Woywodt, 2011).
11 Obviously the percentages did change somewhat. In Slovakia for example 69% of the population was
estimated to be Roman Catholic in 2001 as opposed to 62% in 2011 (UN Data, n.d.). We could expect that most countries see a decline in Roman Catholicism and an increase in non-religious people. As a result estimated coefficients on religious beliefs might be inflated.
10 The third and final regression added a set of dummy variables representing different regions: Europe12, Latin America, the Middle East13, and Others. The inclusion of regions controlled for confounding factors related to unobserved heterogeneity across country groups.
12 Europe is divided and further classified as: Ireland-UK, the Netherlands-Belgium, Germany-Austria-Hungary,
Lithuania-Latvia, Poland-Slovakia, Bulgaria-Romania-Croatia-Slovenia, and Spain-Greece (Ugur, 2015, p. 1568).
13 Tunisia was added to the Middle East because its demographics, mainly religious, were most alike to that
11 III. Results
Figure 2 does not show a clear difference between average family refusal rates in informed and presumed consent countries. However, certain patterns become apparent after categorising countries per region. In Figure 3 we see that both in Europe and Latin America the presumed consent system is predominant. In Europe countries with this system have an overall lower family refusal rate than informed consent countries. In Latin America the average refusal rate seems to be higher than in Europe, but not as high as in the Middle East and Tunisia. Switzerland, Australia, and the USA all have an informed consent system and, with the exception of the USA, average rates similar to those of European countries with the same system. The large differences between countries with similar consent across regions point towards heterogeneity. This might be caused by religious differences, distrust in the system, or some factor not addressed in this paper.
Figure 2. Average Family Refusal Rate per country for the period 2003-2015 0,00% 10,00% 20,00% 30,00% 40,00% 50,00% 60,00% 70,00% 80,00% LB TN PE TR NL IL CH AR UK AU SI PA BG BR LT PY LV CO GR RO DE EC CR US A ES IE CU SK BE PL AT HU Av era ge Fami ly Ref u sal Rat e
informed consent countries presumed consent countries
12 Figure 3. Average Family Refusal Rate per region for the period 2003-2015
Table 3 provides summary statistics for the sample. The second column displays an average family refusal rate of 31,3% for the pool of presumed consent countries. The average for informed consent countries, 36,4%, is visible in the third column. The last column indicates that the difference of 5,1% between these two groups is significant at α=0.05.
For religious beliefs and the mean corruption perception score we also observe significant differences. In the rows of religious beliefs, this is the case for the percentages of the population that are Roman Catholic and non-religious. The proportion of Roman Catholics is larger in presumed consent countries, whereas more non-religious people are observed in an informed consent system. Although the Muslim population is on average 5% higher in presumed consent countries this difference is not statistically significant. Countries with informed consent are generally perceived as less corrupt than countries with presumed consent. The former have a mean corruption perception score of 6.37, the latter one of 4.54. The gap between the two is highly significant.
0,00% 10,00% 20,00% 30,00% 40,00% 50,00% 60,00% 70,00% 80,00% NL UK SI BG LT LV GR RO DE CR ES IE SK BE PL AT HU PE AR PA BR PY CO EC CU TR LB IL TN CH AU US A Av era ge Fami ly Ref u sal Rat e Europe Latin America
Middle East Other informed consent countries
13 Table 3. Descriptive statistics (means and standard deviations for (2003–2015)
Entire sample Presumed
Consent
Informed Consent
Difference Family Refusal Rate 0.333 (0.20) 0.313 (0.21) 0.364 (0.16) -0.051*(0.025) Religious beliefs % of Roman Catholics % of Muslims % of people having no religion 0.453 (0.32) 0.102 (0.25) 0.110 (0.11) 0.526 (0.32) 0.122 (0.30) 0.090 (0.09) 0.347 (0.28) 0.074 (0.14) 0.140 (0.13) 0.179***(0.03) 0.048 (0.025) -0.05*** (0.011)
Corruption Perception Score 5.284 (2.0) 4.54 (1.53) 6.37 (2.11) -1.83*** (0.179)
Number of countries 32 19 13
* p < .05, ** p < .01, *** p < .001
In Table 4 the regression output for the log of the family refusal rate is summarised. Model (1) only contains a binary variable for presumed consent legislation. According to this model presumed consent countries have a 27% lower family refusal rate than informed consent countries. This is a very large and significant effect, which seems to confirm the hypothesis of Ugur (2015). The next of kin are overall more willing to give consent when the deceased is registered as a donor, even if there is a chance that consent is just presumed.
Model (2) accounts for possible confounding factors by adding control variables for religious beliefs and distrust in the system. This model does a better job explaining the variance in the refusal rate than model (1) does. The coefficient of presumed consent increases to 39%. Both the size of the Roman Catholic and the Muslim part of the population have a significant effect on refusal rates. When the proportion of Muslims is large we observe high refusal rates. For a large share of Roman Catholics it is the other way around. The estimator on Roman Catholicism is in line with the general attitude of its church towards organ donation, which it views as a praiseworthy act of selflessness (Ahmed, Oliver, Saif, & Woywodt, 2011, p. 439). The negative relation between the corruption perception score and the refusal rate confirms that when people have more trust in the system consent is more readily given.
14 Model (3) accounts for country group fixed effects. The corruption score now has a positive coefficient. It is however a small and insignificant one, which might be an indication of the overgeneralisation of the measure used. The most notable difference between models (2) and (3) is the coefficient on presumed consent, it is positive and insignificant. This result suggests that both model (1) and model (2) suffer from omitted variable bias caused by some unobserved heterogeneity across country groups. After controlling for this heterogeneity, the consent system no longer has a significant effect on family refusal rates. The effects of the percentage of Roman Catholics and people without a religion increase drastically. Especially the coefficient of 3.90 seems out of place. If we however note that for most countries with non-religious citizens this group accounts for less than 20% of the population then the effect, though still large, is not as staggering as the coefficient would have one presume.
Table 4. Pooled OLS Estimates of Log Family Refusal Rate
(1) (2) (3) Presumed consent -0.272** (0.083) -0.388*** (0.099) 0.0733 (0.083) Religious beliefs % of Roman Catholics % of Muslims
% of people having no religion
-0.397** (0.128) 0.907*** (0.147) 0.546 (0.448) -0.620*** (0.130) 0.269 (0.203) 3.898*** (0.732)
Corruption Perception Score -0.0981** (0.032) 0.0547 (0.041)
Country Group Fixed Effects - - +
R-squared 0.036 0.232 0.610
N 253 253 253
15 IV. Discussion
Based on the descriptive statistics and the regression output there is little reason to think that family refusal rates are higher under a presumed consent system than under an informed one. In fact, the results in this paper seem to point towards the contrary. Presumed consent countries enjoy somewhat lower refusal rates, although no significant contribution therefore can be assigned to legislation itself. A drawback of this inference is that it is not based on changes observed when countries switched systems. Figure 4 sheds some light on this aspect. The family refusal rate in Venezuela does not portray a noticeable change after the enactment of presumed consent, which is in line with model (3). For Chile on the other hand we see a very pronounced increase in refusal rates. Prior to the change in legislation refusal rates fluctuated between 41% and 32%. From 2011 onwards these percentages went up to 50%. A possible reason for the observed increase could be mistrust towards the health care system in Chile (Domínguez & Rojas, 2013). In the presence of such mistrust it is not unlikely that a change in legislation causes an adverse reaction. This might especially be the case if the change is one from informed - to presumed consent, as such a change is often viewed as a violation of the individual’s autonomy (Gill, 2004).
The effect of the default rule could be at play as well. People in a country where informed consent is the rule might overvalue the perceived loss14 of autonomy resulting from a switch in legislation. A potential benefit might similarly be undervalued. In most countries in Europe15 a majority of people has a positive attitude towards organ donation and is willing to donate one of their organs after death (European Commision, 2010). A majority of the vast pool of unregistered people in informed consent countries would therefore benefit from having their preference aligned with their status in the donor registry; as a result overall welfare would increase. Such an apparent gain might loom small in the face of loss aversion and suspicion of a new system. If a government does not adequately address these issues then donation rates might suffer in the short term.
14 The loss could rightly be called a ‘perceived’ loss, because a presumed consent system does not take away an
individual’s freedom to make a choice (Sunstein & Thaler, 2003)
16 Figure 4. Family refusal rates in Chile and Venezuela before and after switching to presumed consent
IV. A. The Dutch Case
The concerns regarding a change in consent legislation highlight the need for the Dutch government to reassure its citizens and clearly communicate its reasons for deciding upon such a change. The government seems to be aware hereof and addresses the issue by sending all eligible non-registered citizens a letter to incentivize them to make an active decision. After six weeks a reminder is sent in the case the individual has still not registered a choice. Only when someone has not taken any action within six weeks after the reminder will he be registered16 as having ‘no objection’ to donating (S. 33 506, 2016).
The ‘no objection’ category in the Dutch case is the most notable deviation from standard presumed consent legislation. The result of this inclusion is that presumed and explicit donors will be clearly differentiated in the registry. Under standard presumed consent everyone that is not a non-donor is a donor. There is no way to tell people that would have been explicit donors under informed consent apart from the presumed ones. The findings in this paper do not support the claim that presumed consent countries experience higher family refusal rates because of a decrease in the explicitness of the consent to donate by the deceased. However, if a presumed consent system maintains the same degree of explicitness as an informed consent system family refusal rates will probably remain unaltered. Still,
16 A confirmation of the registration will be sent within six weeks.
0,00% 10,00% 20,00% 30,00% 40,00% 50,00% 60,00% 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 Fami ly Re fu sal Rat e Chili Venezuela First year under presumed consent system
17 refusal rates and the number of potential donors could be affected in a different way. To capture the effects of a change in legislation a simple formula for the deceased donor rate is useful:
(2) 𝐷𝑒𝑐𝑒𝑎𝑠𝑒𝑑 𝐷𝑜𝑛𝑜𝑟 𝑅𝑎𝑡𝑒 𝑝𝑚𝑝 = 𝑃𝑜𝑡𝑒𝑛𝑡𝑖𝑎𝑙 𝐷𝑜𝑛𝑜𝑟𝑠 𝑝𝑚𝑝 × (1 − 𝐹𝑎𝑚𝑖𝑙𝑦 𝑅𝑒𝑓𝑢𝑠𝑎𝑙 𝑅𝑎𝑡𝑒) + 𝜀
where, pmp = per million people17
Potentials donors = the sum of individuals medically fit for donation and not registered as non-donor = the sum of individuals for whom family consent is sought
ε = error term = the potential donors for whom, for whatever reason, no consent is sought
This equation however fits a standard presumed consent system better than the proposed system in the Netherlands, because as with informed consent the next of kin will know whether the deceased is an explicit donor or not. A slightly altered formula suitable for both an informed consent as well as the proposed system is:
(3) 𝐷𝑒𝑐𝑒𝑎𝑠𝑒𝑑 𝐷𝑜𝑛𝑜𝑟 𝑅𝑎𝑡𝑒 𝑝𝑚𝑝
= [𝑃𝑜𝑡𝑒𝑛𝑡𝑖𝑎𝑙 𝐷𝑜𝑛𝑜𝑟𝑠1 𝑝𝑚𝑝 × (1 − 𝐹𝑎𝑚𝑖𝑙𝑦 𝑅𝑒𝑓𝑢𝑠𝑎𝑙 𝑅𝑎𝑡𝑒1)] + [𝑃𝑜𝑡𝑒𝑛𝑡𝑖𝑎𝑙 𝐷𝑜𝑛𝑜𝑟𝑠2 𝑝𝑚𝑝 × (1 − 𝐹𝑎𝑚𝑖𝑙𝑦 𝑅𝑒𝑓𝑢𝑠𝑎𝑙 𝑅𝑎𝑡𝑒2)] + 𝜀
where, Potential Donors = The sum of people medically fit for donation Potential Donors1 = Potential donors registered as donor
Potential Donors2 = Potential donors not registered as a donor nor as a non-donor
Family Refusal Rate1 > Family Refusal Rate218
17 This is the standard way to denote deceased donor rates in literature.
18 Data reveal refusal rates in the Netherlands of less than 10% if the deceased is a registered donor as opposed
18 Equation (3) should hold true for the Netherlands both before and after the enactment of the new system because family is always approached to ask for consent unless the deceased is a registered non-donor (Nederlandse Transplantatie Stichting).
Based on equation (3) the deceased donor rate is positively affected by an increase in either Potential Donors1 or Potential Donors2, or by a decrease in either Family Refusal Rate1 or Family Refusal Rate2. Both Potential Donors1 and Potential Donors2 should increase if the medical procedures surrounding the act of transplantation are improved: the spotting of eligible donors, the preservation of donors, etc. The change in the Netherlands concerns just the registration method, leaving the medical procedures unaltered. The deceased donor rate is therefore not expected to increase through this channel.
Nothing points towards family refusal rates decreasing because of the properties of the new consent system in the Netherlands. They might however decrease if family discussion on organ donation is boosted as a result of the change. The effects of discussion will be largest when a majority of the population wants to become a donor after death, which is the case in the Netherlands (van Raaij & Taels, 2009, p. 15; European Commision, 2010, p. 17). A discussion among family members should lead to the surfacing of this opinion and consequently refusal rates should drop. Because a majority has a positive attitude towards donating one’s organs we could also expect a rise in Potential Donors1 larger than a decrease in Potential Donors219 in response to the two letters asking all unregistered citizens to make a choice. The combined result of these partial effects would increase deceased donor rates.
A downside of these effects might be that they are short-term in nature. Debate decreases when the change becomes a thing of the past. Furthermore, previous20 large scale organ donation awareness campaigns by the government failed to produce significant effects on family refusal rates (see Figure 1). There is therefore little reason to be optimistic about the positive effect of the new legislation on the deceased donor rate. Certain however is that the introduction of the new system will come at a cost of around €25 to €35 million (S. 33 506, 2016, p. 10).
19 Potential Donors
2 would decrease because some people will register as non-donor.
20 As part of the Masterplan Orgaandonatie from 2008 until 2014 €10 million was spent on media campaigns to
19 Equation (2) can be extended to most presumed consent countries and equation (3) to most informed consent countries if we stick to the argument that regardless of the registration system in place family can always refuse donation (Boyarsky, et al., 2012; Beitel, et al., 2012). These equations reveal some of the main insights of this study.
The argument by Coppen et al. (2010) that next of kin are less likely to refuse donation when the deceased is an explicit donor than when he is a presumed donor seems to be confirmed. Family Refusal Rate1 is in the Netherlands below 10% (Nederlandse Transplantatie Stichting). Figure 2 shows that this refusal rate is lower than that of almost all presumed consent countries.
The argument that family refusal rates are higher if a deceased is not registered compared to when a deceased is a presumed donor also seem to hold true based on data from the Netherlands. Refusal rates when the deceased is not registered are almost 70% and thus higher than those of almost all presumed consent countries (Nederlandse Transplantatie Stichting).
Although the confirmation of both argument might seem conflicting it is not. The overall family refusal rate for informed consent countries is the weighted sum of Family Refusal Rate1 and Family Refusal Rate2. In Table 3 this overall refusal rate was observed to be somewhat higher than the refusal rate of the presumed consent countries. This might support the conclusion that the positive effect on refusal rates caused by lower explicitness is simply not enough to offset the negative effect on refusal rates caused by the higher explicitness of all people that under presumed consent are donors and under informed consent are not registered. The negative estimates for presumed consent in model (1) and (2) in Table 4 are in line with this. Model (3) however shows that the remaining difference in refusal rates cannot be contributed to the consent legislation, but rather to some unobserved differences between countries that were not found in this paper.
A last and very important insight of both equations is that deceased donor rates can only be influenced by some factor that either affects potential donors or family refusal rates. The consent system does not influence the pool of potential donors. In both systems the deceased medically fit for donation are treated as potential donors unless they are registered as non-donors. If consent legislation can also not explain the difference between the refusal rates for both systems then there is reason to doubt the results of earlier research that found that presumed consent legislation has a positive effect on deceased donor rates. These studies
20 might suffer from upwards biased estimates, because refusal rates are negatively correlated with both presumed consent (see Appendix) and deceased donor rates.
V. Concluding Remarks
The main aim of this paper was to address the impact of organ donation consent legislation on family refusal rates. The next of kin often have the final say in the organ procurement process of the deceased. Therefore, family refusal rates are a main determinant of deceased donor rates. Using a sample of 32 countries with a 3:2 ratio of presumed consent countries to informed consent countries a pooled OLS model was used to estimate the effect of consent legislation on refusal rates during the period 2003-2015. The results indicated that countries with presumed consent on average enjoy somewhat lower refusal rates than countries with informed consent. However, after controlling for country group fixed effects the legislative default had an insignificant effect on family consent. The panel data used suffered from two limitations: heterogeneity and being unbalanced. Both could cause the estimator to produce biased and inefficient estimates of the variables of interest. No apparent reason was found to think that the unbalanced panel led to bias. Robust standard errors were used to address heterogeneity. This does not however rule out the possibility of inefficient estimates.
Previous literature finds a positive effect of presumed consent legislation on deceased donor rates. Based on the results in this paper the estimated positive coefficient might suffer from omitted variable bias. Similar research should be carried which includes actual data on refusal rates to account for this bias and see if consent still has a significant effect on donation rates. These results might point out fallacies of this study.
21 Appendix
Table 1 Countries included into the analyses
Country Consent System Source of consent
information
Argentina (AR) Presumed 1,2,3,4
Australia (AU) Informed 1,2,4,5
Austria (AT) Presumed 1,2,4,5
Belgium (BE) Presumed 1,2,4,5
Brazil (BR) Informed 1,2,3,4
Bulgaria (BG) Presumed 2,4,5
Colombia (CO) Presumed 1,2,3,4
Croatia (CR) Presumed 1,2,4,5
Cuba (CU) Informed 1,2,3
Ecuador (EC) Presumed 1,2,3
Germany (DE) Informed 1,2,4,5
Greece (GR) Presumed 2,4,5
Hungary (HU) Presumed 1,2,4,5
Ireland (IE) Informed 2,4,5
Israel (IL) Informed 2
Latvia (LV) Presumed 2,4,5
Lebanon (LB) Informed 1
Lithuania (LT) Informed 1,2,4,5
The Netherlands (NL) Informed 1,2,4,5
Panama (PA) Presumed 2,4
Paraguay (PY) Presumed 1,3
Peru (PE) Presumed 3
22
Romania (RO) Informed 1,2,4,5
Slovakia (SK) Presumed 1,2,4,5
Slovenia (SI) Presumed 1,4,5
Spain (ES) Presumed 1,2,4,5
Switzerland (CH) Informed 1
Tunisia (TN) Presumed 1,2
Turkey (TR) Presumed 4,5
UKa Informed 1,2,4,5
USA Informed 1,2,4,5
a As of now the UK should officially be classified as a mixed consent system since Wales adopted presumed consent in 2015.
However, the period under research does not include any years after 2015. Therefore, the UK remains categorised as an informed consent system in this paper.
Table 2 Countries excluded from the analyses Reason
Mixed consent systema:
Estonia, Italy, Norway, Uruguayb
Less than three years of datac:
Venezuela, Mexico, Malaysia, Czech R., Luxembourg, France, Belarus, Costa Rica, Ukraine, Libya, Serbia, Algeria, Bosnia and Herzegovina, Montenegro
Changed legislation in 13 year periodd:
Chile, Venezuela
Too little observations per yeare:
Malta, Cyprus, Iceland, Moldova, Macedonia Inconsistent consent legislationf:
Domincan R.
a Based on a Commission Staff Working Document of the European Commission and (Ferguson, O'Carroll, & Shepherd,
2014). b Although Uruguay has been an informed consent country since 1971, presumed consent exists in the case of a
violent death (Alfonso, et al., 2007) c It has not been checked whether one of these countries could have been excluded for
one of the other reasons. d Based on (Beitel, et al., 2012) and
http://www.eluniversal.com/nacional-y-politica/121126/venezuela-enforces-law-on-default-organ-donation e In these countries the number of interviews
with with family asking for consent often did not exceed 10, in some cases there was just 1 interview. f Switzerland is not
included in this category because in July 2007 a federal law led to the abolishment of different consent systems in different cantons (Beitel, et al., 2012) and the collected data stems from after this period.
23 Correlation between family refusal rates (FRR) and presumed consent (c2)
c2 -0.1289 1.0000 FRR 1.0000 FRR c2 (obs=254) . correlate FRR c2
24 References
[5] Abadie, A., & Gay, S. (2006). The impact of presumed consent legislation on cadaveric organ donation: A cross-country study. Journal of Health Economics, 25(4), 599-629.
Ahmed, A., Oliver, M., Saif, I., & Woywodt, A. (2011). Organ donation, transplantation and religion.
Nephrology Dialysis Transplantation, 2, 437-444. doi: 10.1093/ndt/gfq628
[3] Alfonso, J., Alvarez, I., Andrade, M., Berrios, C., Duque, E., Fajardo, C., . . . Soratti, C. (2007). Organ Donation in Latin America. Transplantation Proceedings, 39(2), 333-335.
[1] Beitel, J., Garg, A. X., Horvat, L. D., Prakash, V., Rosenblum, A. M., & Siminoff, L. A. (2012). The authority of next-of-kin in explicit and presumed consent systems for deceased organ donation: an analysis of 54 nations. Nephrology Dialysis Transplantation, 27(6), 2533-2546. Boyarsky, B. J., Deshpande, N. A., Hall, E. C., Montgomery, R. A., Ros, R. L., Segev, D. L., & Steinwachs,
D. M. (2012, January 27). Potential Limitations of Presumed Consent Legislation.
Transplantation, 93(2), 136-140.
Central Intelligence Agency. (n.d.). The World Factbook. Retrieved January 29, 2017, from https://www.cia.gov/library/publications/the-world-factbook/fields/2122.html
Coppen, R. (2010). Organ donation, policy and legislation: With special reference to the Dutch organ
Donation Act. Tilburg University. Utrecht: Labor Grafimedia BV. Retrieved January 28, 2017,
from https://pure.uvt.nl/ws/files/1249032/Coppen_Organ_030910_emb_tot.pdf
Coppen, R., Friele, R. D., Gevers, S. K., & Marquet, R. L. (2005, November). Opting-out systems: no guarantee for higher donation rates. Transplant International, 18(11), 1275-1279.
Coppen, R., Friele, R. D., Gevers, S. K., & van der Zee, J. (2010, September 20). Imagining the Impact of Different Consent Systems on Organ Donation: The Decisions of Next of Kin. Death Studies,
34(9), 835-847.
Domínguez, J., & Rojas, J. L. (2013, May). Presumed Consent Legislation Failed to Improve Organ Donation in Chile. Transplantation Proceedings, 45(4), 1316-1317.
doi:10.1016/j.transproceed.2013.01.008
Dool, P., & Steenbergen, E. (2016, September 13). Tweede Kamer stemt voor nieuw systeem
orgaandonatie. Retrieved from
https://www.nrc.nl/nieuws/2016/09/13/tweede-kamer-stemt-voor-nieuw-systeem-orgaandonatie-a1521169
European Commision. (2010). Organ donation and transplantation. Eurobarometer, Brussels. Retrieved 25 1, 2017, from
http://ec.europa.eu/public_opinion/archives/ebs/ebs_333a_en.pdf
Faun, H., Hollander, M., Span, Tommy, & van Raaij, W. F. (2014). Registratiesystemen voor
orgaandonatie: Maatschappelijk draagvlak en effectiviteit. Panteia, Zoetermeer.
[2] Ferguson, E., O'Carroll, R. E., & Shepherd, L. (2014). An international comparison of deceased and living organ donation/transplant rates in opt-in and opt-out systems: a panel study. BMC
Medicine, 12, 1-14.
Gill, M. B. (2004). Presumed Consent, Autonomy, and Organ Donation. Journal of Medicine and
25 Goldstein, D. G., & Johnson, E. J. (2004, December 27). Defaults and Donation Decisions.
Transplantation, 78(12), 1713-1716.
Kahneman, D., Knetsch, J. L., & Thaler, R. H. (1991). The Endowment Effect, Loss Aversion, and Status Quo Bias. The Journal of Economic Perspectives, 5(1), 193-206.
[4] Mcdaid, C., Myers, L., Norman, G., Rithalia, A., Sowden, A., & Suekarran, S. (2009). A systematic review of presumed consent systems for deceased organ donation. Health technology
assessment , 13(26), 1-95.
Nederlandse Transplantatie Stichting. (n.d.). Jaarverslag. Leiden: Nederlandse Transplantatie Stichting. Retrieved January 27, 2017, from
http://www.transplantatiestichting.nl/winkel/jaarverslagen
Newsletter Transplant. (n.d.). Reports and Publications. Retrieved January 24, 2017, from edqm: https://www.edqm.eu/en/organ-transplantation-reports-73.html
NTS. (n.d.). Over de NTS. Retrieved January 21, 2017, from Nederlandse Transplantatie Stichting: http://www.transplantatiestichting.nl/over-de-nts
Stock, J. H., & Watson, M. W. (2015). Introduction to Econometrics (3th ed.). Essex, England: Pearson. Sunstein, C. R., & Thaler, R. H. (2003, October 1). Libertarian Paternalism Is Not an Oxymoron. The
University of Chicago Law Review, 70(4), 1159-1202.
Transparency International. (n.d.). Corruption Perceptions Index. Retrieved January 29, 2017, from https://www.transparency.org/cpi2015/
U.S. Department of State. (n.d.). International Religious Freedom Report for 2015. Retrieved January 29, 2017, from
https://www.state.gov/j/drl/rls/irf/religiousfreedom/index.htm?dlid=148969&year=2010 Ugur, Z. B. (2015, December). DOES PRESUMED CONSENT SAVE LIVES? EVIDENCE FROM EUROPE.
Health Economics, 24(12), 1560-1572.
UN Data. (n.d.). Population by religion, sex and urban/rural residence. Retrieved January 29, 2017, from http://data.un.org/Data.aspx?d=POP&f=tableCode%3A28#POP
van Raaij, W. F., & Taels, M. (2009). Eindrapport: Orgaandonatie, een Vervolgonderzoek naar
Beslissystemen, de Donorkaart en het Donorformulier. Universiteit van Tilburg. Ministerie van
Volksgezondheid, Welzijn en Sport.
Voorstel van wet van het lid Pia Dijkstra tot wijziging van de Wet op de orgaandonatie in verband met het opnemen van een actief donorregistratiesysteem, S. 33 506. (2016)
VWS. (2015, June 10). Beantwoording Kamervragen over effectiviteit donorcampagnes. Retrieved J anuary 28, 2017, from RIjksoverheid:
https://www.rijksoverheid.nl/documenten/kamerstukken/2015/06/10/beantwoording-kamervragen-over-effectiviteit-donorcampagnes
Woolrdige, J. M. (2002). Econometric Analysis of Cross Section and Panel Data. Cambridge, Michigan: MIT Press.
