Sudden cardiac death (SCD) of young people is a devastating but very rare event, with an incidence estimated to be around one per 100,000 person-year in European countries (Reference Wiffen and Clarcke1). To prevent such tragedies, preparticipation screening of young athletes for cardiovascular disease were previously proposed by some European and U.S. scientific associations of cardiologists (Reference Maron, Thompson and Ackerman2;Reference Zipes, Camm and Borggrefe3).
In its most basic format, preparticipation cardiovascular screening consists of taking a personal and family history with a physical (H&P) examination. Some experts also recommend the inclusion of an electrocardiogram (ECG) to increase the sensitivity of such basic examination (Reference Bille, Figueiras and Schamasch4;Reference Corrado, Pelliccia and Bjornstad5) and suggest that screening is cost-effective by selectively referring to some results of economic evaluations without critical evaluation (Reference Corrado, Schmied and Basso6).
However, there is now considerable debate around the benefits and harms of this approach. A recent study (Reference Van Brabandt, Desomer, Gerkens and Neyt7) reported the shortcomings in the evidence to support preparticipation cardiovascular screening in reducing SCD in young athletes. Moreover, according to the World Health Organization, the balance between costs and outcomes must be considered before implementing any screening strategy (Reference Wilson and Junger8).
The aim of this study is, therefore, to perform a systematic review of economic evaluations on preparticipation cardiovascular screening of young athletes to determine whether a preparticipation screening program provides value for money and to highlight the importance of critically assess these economic evaluations before to make a decision based on their conclusions. This study follows a health technology assessment (HTA) report, requested by Belgian authorities and performed by the Belgian Health Care Knowledge Centre, an independent federal institute providing advice to policy makers (Reference Desomer, Gerkens and Vinck9).
METHODS
A systematic search (December 24, 2014) was performed in two stages to identify primary economic evaluations. First, HTA reports were searched by consulting the HTA database of the Cochrane Library and Web sites of HTA agencies. Second, EMBASE, Medline (OVID), Psychinfo (OVID), EconLit (OVID), and NHS EED (National Health Service Economic Evaluation Database of the Cochrane Library) databases were searched to retrieve both reviews of full economic evaluations and primary economic evaluations. Key words and MeSH terms used were related to (i) screening, (ii) SCD or cardiac diseases, (iii) athletes, and (iv) cost and cost-effectiveness. No language restriction was imposed. In the first stage, an HTA report was identified (Reference Wiffen and Clarcke1) performing a search up to January 2013. Because the search strategy in this report respected the criteria for comprehensive research described in the AMSTAR checklist (10) (at least two electronic sources should be searched, the report must include years and databases used, . . .), all primary economic evaluations identified by this report were selected and the search strategy in the second stage was restricted to January 2013 up to December 2014. Full details of the search strategy are available in the appendix of the full HTA report (Reference Desomer, Gerkens and Vinck9).
Predefined selection criteria were applied to retrieve relevant references, in terms of population (young athletes < 35 years old), intervention (a specific cardiovascular preparticipation screening test), comparator (another cardiovascular preparticipation screening test or no screening), and design (full economic evaluations, i.e., studies comparing at least two alternative treatments in terms of costs and outcomes). In terms of outcomes, only the impact on mortality and quality of life was considered relevant. Therefore, only economic evaluations measuring consequences in terms of life-years gained (LYG), life saved (LS), quality-adjusted life-years (QALYs), or disability adjusted life-years are included in this summary.
A two-step selection procedure was performed by two persons. In a first round, references were selected based on title, abstract, and keywords. The second round selection was based on the full text. This procedure was performed by a health economist and medical in- or exclusion criteria were checked by a cardiologist. References of the selected studies were checked for additional relevant citations. Reviews of economic evaluations were used to identify additional primary economic evaluations. Selected primary economic evaluations were then summarized by a health economist in an in-house structured data extraction sheet. The quality of studies was analyzed according to guidelines for economic evaluations (11;Reference Cleemput, Neyt, Van de Sande and Thiry12). Elements that must be taken into account in economic evaluations of screening programs are summarized in Table 1.
Table 1. Elements That Should Be Taken into Account in Economic Evaluations of Screening Programs
Source: adapted from ANAES 2004 (11) and Cleemput 2012 (Reference Cleemput, Neyt, Van de Sande and Thiry12).
RESULTS
The flow chart in Figure 1 describes the results of this search strategy. Seven full economic evaluations were retrieved (Reference Fuller13–Reference Wheeler, Heidenreich, Froelicher, Hlatky and Ashley19). Among these studies, two only reported a cost per case detected (Reference Malhotra, West and Dent16;Reference Menafoglio, Di Valentino and Segatto17). Because this outcome was not considered as relevant, these two studies are not further analyzed in this study but details can be found in the full HTA report. Finally, five economic evaluations are critically assessed. An overview of all input variables and results are provided in the next paragraphs and are critically assessed in the discussion.
Figure 1. Flow chart of the systematic review.
General Information
The economic evaluations were performed for the United States. Only two studies performed a cost-utility analysis (Reference Schoenbaum, Denchev, Vitiello and Kaltman18;Reference Wheeler, Heidenreich, Froelicher, Hlatky and Ashley19). In other analyses, outcomes were assessed in terms of cost per LYG (Reference Fuller13;Reference Leslie, Cohen and Newburger15;Reference Wheeler, Heidenreich, Froelicher, Hlatky and Ashley19) or cost per LS (Reference Halkin, Steinvil, Rosso, Adler, Rozovski and Viskin14). Three studies seemed to adopt a healthcare payer's perspective (not clearly specified) (Reference Fuller13–Reference Leslie, Cohen and Newburger15), while the two other studies adopted a societal perspective and took travel time and cost or patient time into account (Reference Schoenbaum, Denchev, Vitiello and Kaltman18;Reference Wheeler, Heidenreich, Froelicher, Hlatky and Ashley19). Productivity losses were never considered.
All studies also adopted either a lifetime horizon(13;15;18;19) or a period of 20 years (Reference Halkin, Steinvil, Rosso, Adler, Rozovski and Viskin14). While equal discounting at 3 percent has been recommended by the U.S. Panel on Cost Effectiveness in Health and Medicine (Reference Siegel, Torrance, Russell, Luce, Weinstein and Gold20), variations in the discount rate used can be found between studies: 3 percent for both costs and benefits in two studies (Reference Schoenbaum, Denchev, Vitiello and Kaltman18;Reference Wheeler, Heidenreich, Froelicher, Hlatky and Ashley19), 3 percent only for costs in one study (Reference Leslie, Cohen and Newburger15), and no discount rate was applied in two studies (Reference Fuller13;Reference Halkin, Steinvil, Rosso, Adler, Rozovski and Viskin14). Moreover, no sensitivity analyses on changing the discount rate was reported.
Population
All studies focused on competitive (nonprofessional) young athletes in the U.S. population. The exact definition of this population was not always clear. Population was aged between 12 and 35 years old in one study (Reference Halkin, Steinvil, Rosso, Adler, Rozovski and Viskin14), 14 years old in two studies (Reference Schoenbaum, Denchev, Vitiello and Kaltman18;Reference Wheeler, Heidenreich, Froelicher, Hlatky and Ashley19), and 16 years old in one study (Reference Leslie, Cohen and Newburger15). In the latest study, the age was not specified but it concerned high school athletes (Reference Fuller13).
The prevalence of heart diseases in the studied population varied from 0.01 percent (Reference Fuller13) to 1.2 percent (Reference Wheeler, Heidenreich, Froelicher, Hlatky and Ashley19). These variations could in part be explained by the fact that these economic evaluations did not take the same heart diseases into account, that is, two studies only considered heart diseases at risk of SCD (Reference Fuller13;Reference Schoenbaum, Denchev, Vitiello and Kaltman18) while others considered every heart disease (Reference Halkin, Steinvil, Rosso, Adler, Rozovski and Viskin14;Reference Wheeler, Heidenreich, Froelicher, Hlatky and Ashley19) or most important heart diseases (Reference Leslie, Cohen and Newburger15).
Intervention and Comparator
Screening strategies investigated are described in Table 2. To be able to construct an efficiency frontier, all relevant strategies should be compared (Reference Neyt and Van Brabandt21). Nevertheless, disagreements between studies were found concerning the identification of relevant strategies. Some studies did not consider a “no screening” strategy because they argue the current practice is H&P. Some studies also considered a screening with ECG alone, while others notice that in practice, an ECG was always accompanied by H&P. Only the study of Halkin et al. (Reference Halkin, Steinvil, Rosso, Adler, Rozovski and Viskin14) assessed the impact of a yearly screening while the other analyses focused on a one-time screening.
Table 2. Description of Interventions Investigated
a In the Supplementary Materials of the article.
b ECG only for negative H&P.
c The decision is based on ECG results alone, but the cost of H&P is included.
ECG, rest-electrocardiogram; Echo, echography; H&P, history and physical examination; (x), investigated; (-), not investigated.
Patient Pathway and the Performance of Screening Tests
Variations concerning the performance of screening tests were found (see Supplementary Table 1). After a positive result, the patient pathway was rather unclear and diagnostic tests identified varied across studies: family testing with ECG, echography, exercise stress test, Holter monitor, blood pressure monitoring, magnetic resonance imaging, genetic testing, cardiac electrophysiology study, drug studies, catheterization, and coronary angiography. At the end of this second round, the percentage of people with a suspected heart disease also varied, from 0.1 percent (Reference Fuller13) to 2 percent (Reference Halkin, Steinvil, Rosso, Adler, Rozovski and Viskin14).
For people with a suspected heart disease, variations in treatments and in the patient repartition between these treatments were also found across studies (medication, ablation, cardiac surgery, implantable cardioverter defibrillators). Often, only a simple follow-up was assumed.
Quality of Life Data
In the two studies that included the impact on the quality of life, an important part of utility values were based on expert opinions and assumptions because no data were available (see Supplementary Table 2). The impact of sport disqualification or sport restriction was only partially taken into account in one study (Reference Schoenbaum, Denchev, Vitiello and Kaltman18).
Other Important Input Parameters for Outcomes
The risk of SCD in athletes and the effectiveness of patients’ management were also important parameters and variations in estimates were found (see Supplementary Table 3). These estimates were either based on assumptions (Reference Fuller13) or on an Italian study (Reference Corrado, Basso, Pavei, Michieli, Schiavon and Thiene22;Reference Corrado, Basso, Rizzoli, Schiavon and Thiene23). Even if the same source was used (i.e., the Italian study) SCD risks for people with a heart disease differed because the definition of considered heart diseases was heterogeneous between studies (as stated above).
Table 3. Summary of Results of the Economic Evaluations
a Own calculation;
b From probabilistic analyses, based on the median.
H&P, history and physical examination; ECG, rest-electrocardiogram; QALY, quality-adjusted life-year; LS, life saved; LYG, life years gained.
Cost Parameters
Cost data are all based on U.S. data, which limit the transferability to other countries. Cost of screening tests varied between studies (see Supplementary Table 4), with a range of 0(13) USD - 112(18) USD for H&P; 107(19) USD - 360(14) USD for H&P + ECG; and 10(13) USD - 155(18) USD for ECG alone). In one study (Reference Fuller13), the cost of screening tests were particularly low compared with other studies (e.g., 10 USD for an ECG).
Two studies (Reference Fuller13;Reference Halkin, Steinvil, Rosso, Adler, Rozovski and Viskin14) only took into account the cost of screening tests and of the second round, while others also included the management cost of patients with a heart disease (Reference Leslie, Cohen and Newburger15;Reference Schoenbaum, Denchev, Vitiello and Kaltman18;Reference Wheeler, Heidenreich, Froelicher, Hlatky and Ashley19).
Assessment of Uncertainty
Parameter and structural uncertainty were not tackled in two studies (Reference Fuller13;Reference Halkin, Steinvil, Rosso, Adler, Rozovski and Viskin14). Two other studies performed both a one-way sensitivity analysis and a probabilistic sensitivity analysis (Reference Schoenbaum, Denchev, Vitiello and Kaltman18;Reference Wheeler, Heidenreich, Froelicher, Hlatky and Ashley19). Leslie et al. (Reference Leslie, Cohen and Newburger15) used a best case scenario as reference and limited the handling of uncertainty to a one-way sensitivity analysis.
Summary of Results
A synthesis of the results of the economic evaluations is presented in Table 3.
H&P
Compared with other screening strategies, and by taking the “no screening” option into account, H&P alone was either dominated or extendedly dominated in all studies (Reference Fuller13;Reference Wheeler, Heidenreich, Froelicher, Hlatky and Ashley19).
H&P + ECG, Compared with no Screening
Compared with no screening, a screening with H&P and ECG was not considered as a cost-effective strategy in all studies.
The probabilistic analysis performed by Wheeler et al. (Reference Wheeler, Heidenreich, Froelicher, Hlatky and Ashley19) showed that the probability of being cost-effective is 0 percent at a willingness to pay threshold of 50,000 USD/LYG. Moreover, by taking into account the quality of life, a screening with ECG and H&P could even be a dominated strategy, as shown by the 95 percent confidence interval (CI) (see Table 3). According to univariate sensitivity analyses, results were mostly sensitive to the frequency of screening (i.e., an annual screening was unlikely to be cost-effective), the prevalence of the disease (e.g., if the prevalence of heart disease was 0.3 percent instead of 1.2 percent, the incremental cost-effectiveness ratio [ICER] would become close to 200,000 USD/LYG instead of 76,100 USD/LYG), the risk of SCD in athletes compared with nonathletes, the cost of screening tests, and the specificity of screening tests (Reference Wheeler, Heidenreich, Froelicher, Hlatky and Ashley19).
In the study of Leslie et al. (Reference Leslie, Cohen and Newburger15), authors concluded that a screening with H&P and ECG was not cost-effective at a willingness to pay threshold of 50,000 USD/LYG using optimistic parameters. Because this was the result of a “best case” analysis, they argue that such a conclusion can be maintained in the real world.
Halkin et al. (Reference Halkin, Steinvil, Rosso, Adler, Rozovski and Viskin14) assessed a yearly screening with H&P and ECG and concluded that such a screening resulted in an enormous cost per LS, that is, 10,600,000 USD per LS. Under the (overoptimistic) assumption that people could live an additional 100 years, this would lead to an undiscounted ICER of 106,000 USD/LY.
H&P + ECG, Compared with H&P
If H&P is considered as the current standard practice, results are slightly more positive. Wheeler et al. (Reference Wheeler, Heidenreich, Froelicher, Hlatky and Ashley19) concluded that compared with H&P alone, adding an ECG might be considered as cost-effective, depending of the accepted willingness-to-pay threshold, with ICERs of 42,100 USD/LYG (95 percent CI, 21,200 USD – 71,300 USD) and 61,600 USD/QALY gained (95 percent CI, 15,300 USD-dominated). When taking into account the possible impact on quality of life, the CI from the probabilistic analysis showed that adding an ECG could be a dominated strategy (see Table 3) (Reference Wheeler, Heidenreich, Froelicher, Hlatky and Ashley19).
Schoenbaum et al. (Reference Schoenbaum, Denchev, Vitiello and Kaltman18) showed that compared with H&P alone, adding an ECG suggested an ICER of 68,800 USD/QALY gained, with a probability to be cost-effective at a threshold of 50,000 USD/QALY gained of 30 percent. Given these results, they concluded that a screening with H&P and ECG was not a cost-effective strategy compared with H&P alone. The univariate sensitivity analysis also showed that results were sensitive to the relative risk of SCD for athletes compared with nonathletes, the prevalence of heart disease in the population, the specificity of ECG and H&P and the cost of ECG.
ECG Alone, Compared with no Screening
Compared with no screening, Wheeler et al. (Reference Wheeler, Heidenreich, Froelicher, Hlatky and Ashley19) obtained an ICER above 50,000 USD/QALY. Moreover, the credibility intervals showed that by taking into account the impact on the quality of life, ECG alone could even be a dominated strategy (see Table 3).
ECG Alone, Compared with H&P
If H&P is considered as the current standard practice, results are more positive. Compared with H&P, Fuller (Reference Fuller13) obtained an ICER of 44,000 USD/LYG. No sensitivity analysis was performed. Schoenbaum et al. (Reference Schoenbaum, Denchev, Vitiello and Kaltman18) also compared ECG alone with H&P. They obtained an ICER of 37,700 USD/QALY gained and a probability to be cost-effective at a threshold of 50,000 USD/QALY of 66 percent. Univariate sensitivity analysis showed that results were highly sensitive to the relative risk of SCD for athletes compared with nonathletes, the prevalence of heart disease in the population, and the specificity of ECG and H&P.
Echo Alone
Only one study (Reference Fuller13) assessed the impact of performing an echography alone during the first screening round and showed that such a strategy was not cost-effective (no sensitivity analysis was performed).
DISCUSSION
The cost of screening seems a small price to save the life of a child. Nevertheless, different arguments need to be analyzed before concluding that preparticipation cardiovascular screening would be a cost-effective strategy. Elements to take into account are the performance of the screening tests to detect people at risk of SCD, the cost and risk associated with further evaluations (including the impact on the quality of life), and the impact of interventions after a confirmed diagnosis.
Results presented in the previous section showed that the cost-effectiveness of a preparticipation cardiovascular screening program for young athletes is quite doubtful. Moreover, there are numerous methodological concerns.
Impact of the Comparator
The three economic evaluations that concluded in favor of a screening (Reference Fuller13;Reference Schoenbaum, Denchev, Vitiello and Kaltman18;Reference Wheeler, Heidenreich, Froelicher, Hlatky and Ashley19), assumed that H&P was the standard of care and a “no screening” option was not taken into account. Nevertheless, there is no evidence showing that H&P should be the current standard practice (Reference Wiffen and Clarcke1;Reference Haskell24). The study of Wheeler et al. (Reference Wheeler, Heidenreich, Froelicher, Hlatky and Ashley19) estimated that this strategy would be extendedly dominated if a no screening strategy is also taken into account. This strategy is clearly a relevant alternative and all economic evaluations including no-screening as a comparator concluded against screening.
Performance of Screening Tests and the Underlying Diseases
Sensitivity analysis showed that the performance, and especially the specificity of screening tests, has an important impact on results. Variations in ECG specificity observed between models could partly be explained by an evolution of diagnostic criteria. Indeed, imperfect clinical data now suggest a specificity of 95 percent if the test is performed by expert sports electrocardiographists using the Seattle criteria (Reference Van Brabandt, Desomer, Gerkens and Neyt7;Reference Drezner, Ackerman and Anderson25;Reference Sharma, Merghani and Gati26). Nevertheless, for a mass screening not always performed by a cardiologists, results might be expected to be worse (Reference Van Brabandt, Desomer, Gerkens and Neyt7). The systematic review of the literature has also been reported that a combination of ECG and H&P might lead to a specificity of 70 percent (Reference Van Brabandt, Desomer, Gerkens and Neyt7;Reference Magalski, McCoy and Zabel27), which is below values used in the models.
Moreover, most studies (Reference Fuller13;Reference Leslie, Cohen and Newburger15;Reference Schoenbaum, Denchev, Vitiello and Kaltman18) did not include all heart diseases that could cause SCD (e.g., Brugada syndrome, etc.) in their model. Seeking for more heart diseases could reduce the specificity of the screening, which will also negatively impact the costs and quality of life (more false positive, over-diagnosis, and over-treatment) making the ICER even more unfavorable.
Risk of Sudden Cardiac Death and the Effectiveness of Patient Management
All studies assumed that the SCD rate was higher in athletes than in nonathlete, which has an important impact on the effectiveness of patient management, that is, the risk reduction of SCD once diagnosed and, for example, disqualified from sport. This artificially increased the impact of screening and the only comparative data supporting a higher risk of SCD in athletes comes from an Italian study (Reference Corrado, Basso, Pavei, Michieli, Schiavon and Thiene22), while other sources show similar or even higher rates in nonathletes than in athletes (Reference Van Brabandt, Desomer, Gerkens and Neyt7;Reference Estes and Link28;Reference Link and Estes29). There is also a lot of disagreement among experts about the management of diagnosed diseases in asymptomatic individuals and it is not known whether avoiding vigorous exercise will prevent SCD to occur in other circumstances (Reference Van Brabandt, Desomer, Gerkens and Neyt7).
Moreover, the findings in this Italian study have never been reproduced by other researchers and the results of this nonrandomized study should be regarded with caution. First, only data from one region (Veneto) have been published, although preparticipation screening is mandatory in Italy nationwide. Second, data collected before 1979 and after 2004 are not published while screening in Italy became mandatory by law in 1971. Data before 1979 would be needed to determine whether the unexpectedly high initial SCD incidence rate in 1979 was no simple random variation, and data after 2004 would allow us to better appreciate the consistency of the initial findings. Third, the 2004 SCD incidence rate in Italy (with a mandatory screening) is similar to the SCD incidence rate in other European countries without screening (Reference Van Brabandt, Desomer, Gerkens and Neyt7).
Furthermore, in the Italian study, the risk reduction of SCD was measured for a 26-year period during which a yearly screening was performed (Reference Corrado, Basso, Pavei, Michieli, Schiavon and Thiene22). Assuming that a one-time screening would have the same effect as a yearly screening (as assumed in all studies except the study of Halkin et al.) (Reference Halkin, Steinvil, Rosso, Adler, Rozovski and Viskin14) is very optimistic. The study of Wheeler et al. (Reference Wheeler, Heidenreich, Froelicher, Hlatky and Ashley19) showed that, if a yearly screening were assumed instead of a one-time screening, all screening strategies were unlikely to be cost-effective.
Impact on the Quality of Life
For people with a (false) positive result, the negative implications on their quality of life was usually not taken into account: long-term side effects of nontrivial therapies and lifetime medication or lifestyle restrictions and anxiety associated with living with a heart disease. A short-term impact of false positive results and of heart disease diagnoses on the quality of life was only included in two studies (Reference Schoenbaum, Denchev, Vitiello and Kaltman18;Reference Wheeler, Heidenreich, Froelicher, Hlatky and Ashley19), and the possible impact of sport disqualification was only partially taken into account in one study (Reference Schoenbaum, Denchev, Vitiello and Kaltman18). No robust data on quality of life seemed to be available, which makes it impossible to correctly assess such an impact.
Nevertheless, the combination of a very low prevalence and a specificity between 70 percent and 95 percent would result in a large proportion of the screened population being confronted with the negative consequences of a screening strategy in asymptomatic young athletes (Reference Van Brabandt, Desomer, Gerkens and Neyt7). In the absence of a perfect diagnostic test, preparticipation cardiovascular screening inevitably induces a huge number of false-positives, leading to overdiagnosis and overtreatment of the “healthiest” segment of the population.
Impact of Costs
Cost estimates used in the studies are based on U.S. data and are expected to be higher than in European countries. Nevertheless, when performed, sensitivity analyses showed that results were not sensitive to cost data, with exception of the screening tests costs for which the negative impact of increasing such a cost was highlighted. It should also be noted that productivity losses were not taken into account in the studies. Nevertheless, even if the weaknesses of clinical data do not allow to estimate the exact impact, it seems that the harms induced by screening might, even in terms of mortality, be larger than the benefits (Reference Van Brabandt, Desomer, Gerkens and Neyt7). Costs are, therefore, only secondary arguments.
Ethical Considerations
It should also be noted that screening of athletes is intrinsically subject to ethical questions because it is a medical approach that aims to discover abnormalities in persons who consider themselves as being in good health. Screening may hamper their freedom of practicing sport and adopting a healthy lifestyle, while there is no clear evidence that such a program on a population level could save lives. This highlights also the importance of the role of the physician who has to inform the athlete on potential consequences of the screening. These and other important ethical considerations are discussed more in detail in the full HTA report (Reference Desomer, Gerkens and Vinck9).
CONCLUSION
The results of this systematic review and critical assessment show that: (i) most studies come up with results not in favor of screening and (ii) those in favor are overoptimistic because, among others, they forget to consider a no-screening strategy compared with which there is no evidence that a mass screening has an added value or they only counted for a one-time screening cost while the underlying study induced yearly costs. A critical assessment of primary economic evaluations is necessary before their conclusions are reported in other studies. Based on current evidence, a mass screening of young athletes for cardiovascular diseases would very probably do more harm than good for the screened population and does not provide value for money. Therefore, introducing or sustaining such a program should be discouraged.
SUPPLEMENTARY MATERIAL
Supplementary Table 1: https://doi.org/10.1017/S0266462317000162
Supplementary Table 2: https://doi.org/10.1017/S0266462317000162
Supplementary Table 3: https://doi.org/10.1017/S0266462317000162
Supplementary Table 4: https://doi.org/10.1017/S0266462317000162
CONFLICTS OF INTEREST
No conflict of interest.
