During the COVID-19 pandemic, far-reaching restrictions on normal life were implemented by laws and regulations to prevent disease transmission and protect people and society. In the absence of evidence for the most appropriate legal mitigation measures, decision makers often cited principles of precaution as the basis for implementing restrictive laws and regulations and unprecedented measures such as lockdowns, curfews, and school closings (1).

To avert harm in crises, many people consider precaution as a morally correct concept for action. Many regard authorities who act with precaution as alert and responsive, fulfilling obligations to take care of people in times of uncertainty. Critics have castigated political leaders who failed to act with caution, particularly when disastrous consequences apparently followed from failure to take strong preventive measures.

Soon after the start of the pandemic, investigators initiated a surge of clinical trials for drugs and vaccines, which have been enormously valuable and contributed to evidence-based improvements in clinical care and disease prevention. However, for public health, such mitigation measures as the 3- and 6-foot distancing rules, strategies for school openings, or quarantine duration remain untested in clinical trials. Their comparative benefit–harm ratio is still guided by the precautionary principle rather than by empirical evidence.

The Precautionary Principle

The precautionary principle was developed to act on environmental risk when there is uncertainty about optimal risk management. In 1992, the United Nations (UN) introduced it as a principle for environmental protection (2). In 1993, the European Union (EU) incorporated the principle into its legislative framework (the Maastricht Treaty).

The 1992 UN declaration did not consider potential harms or burdens of protective measures, but in 2000, the EU Commission introduced evidence-based risk assessment for public health measures under the precautionary principle. It explained that application of the precautionary principle should include a “scientific evaluation, as complete as possible…identifying at each stage the degree of scientific uncertainty” (3).

Evidence-Based Risk Assessment in Times of Crisis

Three factors must be considered in evidence-based risk assessment of public health measures: the absolute magnitude of morbidity and mortality from the threat, the expected benefits of measures against the threat, and the expected harms and burdens of the measures. Often, one or several of these are difficult to quantify, but as a prerequisite for moral action, authorities should explicitly appraise all 3 using the best available evidence (3).

Some mitigation measures against COVID-19, such as improved hand hygiene, have little harm or cost. Others, including closures of schools and workplaces, curfews, or travel restrictions, have serious harms and burdens. Such measures need structured evidence evaluation for all 3 components of evidence-based risk assessment to be recognized.

Against this background, it may not be surprising that on 12 December 2020, the Constitutional Court of Austria declared the country's COVID-19 emergency school closure laws unconstitutional because of a lack of evidence-based risk assessment (4). The court explained that the Austrian government had not provided estimates of expected benefits and harms of school closures. The court did not dismiss school closure as a reasonable public health measure itself but—in keeping with the EU commission (4)—established the need for transparent quantification for proportionality of mitigation measures. On the basis of similar reasoning, the district court of The Hague in February 2021 ruled a Dutch curfew law to be unconstitutional (5).

Evidence Generation in Times of Crisis

In times of crisis, authorities first need to use evidence-based risk assessment to identify knowledge gaps underlying public health interventions introduced by emergency regulations. Next, they must ensure that these gaps are efficiently filled through empirical testing. This is scientifically, morally, and legally superior to the precautionary principle.

Although it remains unthinkable for the U.S. Food and Drug Administration and similar agencies to approve new drugs or vaccines without proper testing in randomized trials, many COVID-19 public health measures are still based on legal process derived through politics, modeling, and expert opinion (1). In some cases, authorities have even actively blocked such evidence generation, as for randomized comparison of school openings (6) and training facilities (7) in Norway. The Norwegian authorities argued that their primary obligation in times of crisis is to protect life and well-being by being cautious, not to facilitate evidence through research.

Admittedly, clinical trials for public health interventions are complex to perform. Political challenges need to be addressed, and stakeholders at the national, regional, and local level (such as schools and parents, or businesses and customers) must be actively involved. But examples such as the Danish randomized trial for face masks against COVID-19 show that such trials are possible and may provide valuable information in crises (2, 8).

Embedded Testing

Because health interventions introduced through laws are binding, evidence generation needs to be embedded in legal implementation and application. Recently, new concepts have been proposed that can be applied to testing of legal mitigation measures (9, 10).

So-called learning health systems can facilitate testing in clinical trials and high-quality observational studies that are seamlessly embedded in the implementation and de-implementation process of legal health interventions. This approach combines best research methodology with concepts of evidence-based policy making.

The concept of learning health systems consists of cycles of comparative testing of interventions. When analyses reveal that one approach is better than the other, the superior strategy is universally adopted. When new policy arises, test cycles start again—for example, by randomization to the current standard for social distancing of 6 feet versus a newer alternative of 3 feet. Cluster randomization of schools, classes, or neighborhoods or stepped-wedge designs with sequential implementation generate valid results. Opt-in and opt-out consent and information strategies are thoroughly considered to ensure ethical testing (10). If randomized testing is not possible, creative new observational study designs, such as causal inference emulation methodology with standardized data collection to simulate clinical trials, may be applied.

Embedded testing allows investigators to find the least restrictive yet effective legal interventions against a new health threat, such as COVID-19. The concept can be applied to many legal health interventions during crises, such as comparing different types of masks or varying strategies for school openings. It allows safer reintroduction of normal business and reduces risk exposure because only persons who are assigned to the new intervention in each testing cycle are at potential risk for incremental harm.

Conclusions

The precautionary principle is a powerful tool that can be used to implement far-reaching restrictions while bypassing standard procedures for legal and medical risk assessment. However, uncritical application of the principle for legal health interventions prevents opportunities to gather new evidence for better handling of future crises. The remarkable drug and vaccine development during the COVID-19 pandemic has shown that rapid generation of high-quality empirical evidence is possible in crises. The same commitment should be given to generating high-quality evidence for legal health interventions and mitigation measures. It will enable decision makers to better quantify expected benefits, harms, and burdens; maintain trust and understanding; and preserve democratic processes.

References

• 1. Burris S ,  Anderson ED ,  Wagenaar AC . The “legal epidemiology” of pandemic control. N Engl J Med. 2021;384:1973-1975. [PMID: 34043295] doi:10.1056/NEJMp2103380 CrossrefMedlineGoogle Scholar
• 2. Bourguignon D; European Parliamentary Research Service. The precautionary principle. Definitions, applications and governance. Publications Office of the European Union; December 2015. Accessed at www.europarl.europa.eu/RegData/etudes/IDAN/2015/573876/EPRS_IDA(2015)573876_EN.pdf on 4 August 2021. Google Scholar
• 3. Commission of the European Communities. Communication from the Commission of the precautionary principle. 2002. Accessed at https://eur-lex.europa.eu/legal-content/EN/TXT/PDF/?uri=CELEX:52000DC0001&from=EN on XXX. Google Scholar
• 4. Verfassungsgerichtshof. VfGH V 436/2020. 10 December 2020. Accessed at www.vfgh.gv.at/downloads/VfGH_10.12.2020_V_436_2020_Covid-Massnahmen_in_Schulen.pdf on 4 August 2021 . Google Scholar
• 5. Rechtbank Den Haag. C/09/607056 / KG ZA 21-118. 16 February 2021. Accessed at https://uitspraken.rechtspraak.nl/inziendocument?id=ECLI:NL:RBDHA:2021:1100&showbutton = true&keyword=avondklok on 4 August 2021. Google Scholar
• 6. Fretheim A ,  Flatø M ,  Steens A , et al. COVID-19: we need randomised trials of school closures [Letter]. J Epidemiol Community Health. 2020;74:1078-1079. [PMID: 32907916] doi:10.1136/jech-2020-214262 CrossrefMedlineGoogle Scholar
• 7. The TRAiN Study Group. Randomized re-opening of training facilities during the COVID-19 pandemic. medRxiv. Preprint posted online 25 June 2020. doi:10.1101/2020.06.24.20138768 Google Scholar
• 8. Bundgaard H ,  Bundgaard JS ,  Raaschou-Pedersen DET , et al. Effectiveness of adding a mask recommendation to other public health measures to prevent SARS-CoV-2 infection in Danish mask wearers: a randomized controlled trial. Ann Intern Med. 2021;174:335-343. [PMID: 33205991] doi:10.7326/M20-6817 LinkGoogle Scholar
• 9. Horwitz LI ,  Kuznetsova M ,  Jones SA . Creating a learning health system through rapid-cycle, randomized testing. N Engl J Med. 2019;381:1175-1179. [PMID: 31532967] doi:10.1056/NEJMsb1900856 CrossrefMedlineGoogle Scholar
• 10. Kalager M ,  Bretthauer M . Improving cancer screening programs. Science. 2020;367:143-144. [PMID: 31919207] doi:10.1126/science.aay3156 CrossrefMedlineGoogle Scholar

中文翻译：

---

合法健康干预的证据和预防措施：从 COVID-19 大流行中学习

在 COVID-19 大流行期间，法律法规对正常生活实施了深远的限制，以防止疾病传播并保护人民和社会。在缺乏最合适的法律缓解措施的证据的情况下，决策者经常引用预防原则作为实施限制性法律法规和前所未有的措施（例如封锁、宵禁和学校停课）的基础 (1)。

为了避免危机中的伤害，许多人认为预防是一种道德上正确的行动概念。许多人认为，采取预防措施的当局是机敏和反应灵敏的，在不确定时期履行照顾人们的义务。批评者谴责了未能谨慎行事的政治领导人，尤其是在灾难性后果显然因未能采取强有力的预防措施而产生的情况下。

大流行开始后不久，研究人员启动了大量药物和疫苗的临床试验，这些试验非常有价值，有助于临床护理和疾病预防的循证改进。但是，对于公共卫生而言，诸如 3 英尺和 6 英尺距离规则、学校开学策略或隔离期等缓解措施仍未在临床试验中进行测试。它们的相对收益-伤害比仍然以预防原则为指导，而不是由经验证据指导。

预防原则

预防原则的制定是为了在最佳风险管理存在不确定性时对环境风险采取行动。1992 年，联合国 (UN) 将其作为环境保护原则 (2)。1993 年，欧盟（EU）将该原则纳入其立法框架（马斯特里赫特条约）。

1992 年的联合国宣言没有考虑保护措施的潜在危害或负担，但在 2000 年，欧盟委员会根据预防原则对公共卫生措施引入了循证风险评估。它解释说，预防原则的应用应包括“科学评估，尽可能完整......在每个阶段确定科学不确定性的程度”（3）。

危机时期的循证风险评估

在公共卫生措施的循证风险评估中必须考虑三个因素：威胁的发病率和死亡率的绝对量级、针对威胁的措施的预期收益以及措施的预期危害和负担。通常，其中一项或多项难以量化​​，但作为道德行为的先决条件，权威机构应使用最佳可用证据明确评估所有 3 项 (3)。

一些针对 COVID-19 的缓解措施，例如改善手部卫生，几乎没有危害或成本。其他措施，包括关闭学校和工作场所、宵禁或旅行限制，都会造成严重的危害和负担。这些措施需要对循证风险评估的所有 3 个组成部分进行结构化的证据评估才能得到认可。

在此背景下，奥地利宪法法院于 2020 年 12 月 12 日宣布该国的 COVID-19 紧急关闭学校法违宪，因为缺乏基于证据的风险评估 (4) 也就不足为奇了。法院解释说，奥地利政府没有提供对关闭学校的预期收益和危害的估计。法院并没有将关闭学校本身视为一项合理的公共卫生措施，但与欧盟委员会 (4) 一致，确定需要对缓解措施的相称性进行透明量化。基于类似的推理，海牙地区法院于 2021 年 2 月裁定荷兰的宵禁法违宪 (5)。

危机时期的证据生成

在危机时期，当局首先需要使用基于证据的风险评估来确定紧急法规引入的公共卫生干预措施背后的知识差距。接下来，他们必须确保通过经验测试有效地填补这些空白。这在科学、道德和法律上都优于预防原则。

尽管美国食品和药物管理局和类似机构在随机试验中未经适当测试就批准新药或疫苗仍然是不可想象的，但许多 COVID-19 公共卫生措施仍然基于通过政治、模型和专家意见得出的法律程序。 1）。在某些情况下，当局甚至积极阻止此类证据的生成，例如对挪威的学校开放 (6) 和培训设施 (7) 进行随机比较。挪威当局辩称，他们在危机时期的主要义务是通过谨慎保护生命和福祉，而不是通过研究促进证据。

诚然，公共卫生干预的临床试验执行起来很复杂。需要解决政治挑战，国家、地区和地方层面的利益相关者（例如学校和家长，或企业和客户）必须积极参与。但丹麦针对 COVID-19 的口罩随机试验等例子表明，此类试验是可能的，并且可能在危机中提供有价值的信息 (2, 8)。

嵌入式测试

由于通过法律引入的卫生干预措施具有约束力，因此需要将证据生成嵌入法律实施和应用中。最近，提出了可用于测试法律缓解措施的新概念 (9, 10)。

所谓的学习卫生系统可以促进临床试验和高质量观察性研究中的测试，这些研究无缝嵌入到合法卫生干预措施的实施和取消实施过程中。这种方法将最佳研究方法与基于证据的政策制定概念相结合。

学习卫生系统的概念包括干预措施的比较测试循环。当分析表明一种方法优于另一种方法时，普遍采用优越的策略。当新政策出现时，测试周期会重新开始——例如，随机化到当前的 6 英尺社交距离标准与 3 英尺的新替代标准。学校、班级或社区的集群随机化或具有顺序实施的阶梯楔形设计会产生有效的结果。选择加入和选择退出同意和信息策略被彻底考虑以确保道德测试 (10)。如果无法进行随机测试，则可以应用创造性的新观察性研究设计，例如具有标准化数据收集的因果推理模拟方法来模拟临床试验。

嵌入式测试使调查人员能够找到限制最少但有效的法律干预措施，以应对新的健康威胁，例如 COVID-19。该概念可应用于危机期间的许多合法健康干预措施，例如比较不同类型的口罩或不同的学校开学策略。它允许更安全地重新引入正常业务并减少风险暴露，因为只有在每个测试周期中被分配到新干预的人员才有潜在的增加伤害风险。

结论

预防原则是一种强大的工具，可用于实施影响深远的限制，同时绕过法律和医疗风险评估的标准程序。然而，对合法健康干预原则的不加批判的应用阻碍了收集新证据以更好地处理未来危机的机会。COVID-19 大流行期间显着的药物和疫苗开发表明，在危机中可以快速生成高质量的经验证据。应该给予同样的承诺，为合法的健康干预和缓解措施提供高质量的证据。它将使决策者能够更好地量化预期的收益、危害和负担；保持信任和理解；并维护民主进程。

参考

• 1. 伯里斯 ,  安德森教育署 ,  瓦格纳尔空调 . 大流行控制的“法律流行病学”。 N Engl J Med。 2021 年；384：1973-1975。[PMID：34043295] 做：10.1056/NEJMp2103380 CrossrefMedlineGoogle Scholar
• 2.布吉尼翁 D; 欧洲议会研究服务。预防原则。定义、应用和治理。欧盟出版办公室；2015 年 12 月。2021 年 8 月 4 日访问 www.europarl.europa.eu/RegData/etudes/IDAN/2015/573876/EPRS_IDA(2015)573876_EN.pdf。Google Scholar
• 3.欧洲共同体委员会。委员会关于预防原则的通知。2002 年访问 https://eur-lex.europa.eu/legal-content/EN/TXT/PDF/?uri=CELEX:52000DC0001&from=EN on XXX。谷歌学术
• 4. Verfassungsgerichtshof。VfGH V 436/2020。2020 年 12 月 10 日。2021 年 8 月 4 日在 www.vfgh.gv.at/downloads/VfGH_10.12.2020_V_436_2020_Covid-Massnahmen_in_Schulen.pdf 访问。谷歌学术
• 5. Rechtbank Den Haag。C/09/607056 / KG ZA 21-118。2021 年 2 月 16 日。访问 https://uitspraken.rechtspraak.nl/inziendocument?id=ECLI:NL:RBDHA:2021:1100&showbutton = true&keyword=avondklok 于 2021 年 8 月 4 日。Google Scholar
• 6. 弗雷特海姆 A ,  弗拉特米 ,  斯蒂恩斯 , 等. COVID-19：我们需要关闭学校的随机试验 [Letter]。 J 流行病社区健康。 2020 年；74：1078-1079。[PMID：32907916] 做：10.1136/jech-2020-214262 CrossrefMedlineGoogle Scholar
• 7. TRAiN 研究组。在 COVID-19 大流行期间随机重新开放培训设施。医学Rxiv。预印本于 2020 年 6 月 25 日在线发布。doi:10.1101/2020.06.24.20138768 Google Scholar
• 8. 邦加德 ,  邦嘉JS ,  拉舒-佩德森 DET , 等. 将口罩建议添加到其他公共卫生措施以预防丹麦口罩佩戴者感染 SARS-CoV-2 的有效性：一项随机对照试验。 安实习医生。 2021 年；174：335-343。[PMID：33205991] 做：10.7326/M20-6817 链接谷歌学术
• 9. 霍维茨李 ,  库兹涅佐娃 ,  琼斯公司 . 通过快速循环、随机测试创建学习型健康系统。 N Engl J Med。 2019 年；381：1175-1179。[PMID：31532967] 做：10.1056/NEJMsb1900856 CrossrefMedlineGoogle Scholar
• 10. 卡拉格 ,  布雷特豪尔 . 改进癌症筛查计划。 科学。 2020 年；367：143-144。[PMID：31919207] 做：10.1126/science.aay3156 CrossrefMedlineGoogle Scholar