一个研究团队分析为什么美国的新冠死亡病例依旧被低估
自2020年3月新冠疫情被宣布构成全球大流行以来，有一个话题不断引起争论，那就是美国官方公布的新冠死亡人数统计数据，是否准确体现了与引发新冠的SARS-CoV-2病毒有关的死亡人数。
有些政客和许多公共卫生从业者认为，新冠死亡人数被高估。例如，2023年1月，《华盛顿邮报》发表的一篇评论文章称，新冠死亡人数不仅包括死于新冠的人数，还包括死于其他原因但恰好感染新冠的人数。
但大多数科学家认为，新冠死亡人数被低估，因为这些数据没有包含被错误分类为其他死因的新冠死亡人数。
我们是由波士顿大学（Boston University）、明尼苏达大学（University of Minnesota）、加州大学旧金山分校（University of California San Francisco）等机构组成的研究团队的成员，自疫情爆发以来持续跟踪新冠死亡人数。团队的主要目标是评估美国新冠死亡人数是否被低估，以及哪些地区的死亡人数被低估。
研究超额死亡人数
研究这个问题的途径之一是分析人口健康研究人员所说的超额死亡率。这个指标将疫情期间的死亡人数与根据疫情之前的趋势预测的死亡人数进行对比。
超额死亡率评估了直接死于新冠的人数，或通过间接途径死亡的人数，例如在新冠高峰期间避免前往医院的患者。确定死因是一个复杂的过程，而记录是否死亡却非常直接。因此，计算超额死亡人数被认为是对疫情死亡人数偏差最小的估算。
一般而言，如果新冠死亡人数超过超额死亡人数，这意味着新冠死亡人数可能被高估，当然有一些重要的说明，会在下文一一解释。如果超额死亡人数超过新冠死亡人数，后者可能被低估。
在最近发布的一项尚未经过同行审议的研究中，我们的团队发现，在2020年3月至2022年2月疫情的前两年，美国超额死亡人数为996,869至1,278,540人。其中866,187人的死亡证明中确认的死因为新冠。这意味着超额死亡人数比新冠死亡人数多130,682至412,353人。疫情第一年和第二年，两者之间存在巨大差距。这意味着即使在经过疫情最初几个月的混乱之后，新冠疫情死亡人数依旧被低估。
多项重要研究也认为，在新冠疫情前两年，全美超额死亡人数超过了新冠死亡人数。我们的团队开展的初步分析发现，超额死亡人数和新冠死亡人数的差距延续到了疫情第三年。这表明，新冠死亡人数现在依旧被低估。
理解数据差异
解释超额死亡人数与公布的新冠死亡人数之间的差异是一项更具有挑战性的任务。但有许多证据证明，数据的差异基本代表未被统计的新冠死亡人数。
我们在最近的一项研究中发现，超额死亡人数在官方公布的新冠死亡人数之前达到最高峰。即使与阿尔兹海默症等死因有关的超额死亡人数也是如此，因为在疫情期间，由于患者避免去医院或其他行为变化，这类超额死亡人数不太可能快速发生变化。
这一研究结果与我们的观察结果一致。我们发现，在新冠高峰开始时，新冠死亡病例可能未被确认或被错误划分为其他死因。目前，医疗提供商和死亡调查员在社区中进行新冠检测的频率可能降低。如果超额死亡的死因并非新冠疫情，则在新冠疫情高峰时期，超额死亡人数应该维持相对稳定，或者在疫情高峰过后，由于医院人满为患和医疗中断导致死亡，从而使超额死亡人数达到最高峰。
疫情期间，与服药过量等外部死因有关的超额死亡人数同样增多。然而，一项初步研究发现，与超额死亡人数的整体增长幅度相比，这类死亡人数的增长幅度相对较小。因此，外因死亡人数也不能解释超额死亡人数和新冠死亡人数之间的差异。
鉴于前文提到的《华盛顿邮报》发表的那篇重要的评论文章，这个证据值得考虑。文章提出，美国的新冠死亡人数被严重高估。作者主张，在某些医院，大范围新冠检测导致感染新冠但死于其他病因的患者，在死亡证明中依旧将新冠作为死因之一。然而，这是一种根本性的误解，将院内死亡人数泛化到全国。
这种过度泛化存在缺陷的一个原因是，院内死亡不同于院外死亡。在医院以外，通常没有新冠检测，死亡调查人员没有接受充分培训，而且对死者不够了解。事实上，我们的研究表明，在医院外，新冠死亡人数被严重低估。
农村地区的验尸官调查报告也显示，在院外死因确认方面存在巨大差异。如果与验尸官自己的政治理念不一致，或者死者家属希望不要提及新冠，有些验尸官甚至会在死亡记录中说明死者未感染新冠。
过度泛化的另外一个问题是地理差异。初步研究表明，美国绝大多数县的超额死亡人数超过新冠死亡人数。尤其是在南方各县、落基山脉各州和农村地区，超额死亡人数超过新冠死亡人数的比例更高。这表明，在这些地区新冠死亡人数可能被低估。
我们的分析结果，至少在非常有限的程度上，可以支持新冠死亡人数有时候被高估的观点。在新英格兰和大西洋中部各州的多个大中型城区，新冠死亡人数超过了超额死亡人数。但美国大多数地区并没有遵循少数几个县的模式。
在新英格兰和大西洋中部各州，将新冠划定为死因的部分死亡人数可能实际上并非死于新冠，对于这种情况可能有其他解释。首先，在这些地区，新冠疫情防控措施，通过与新冠无关的途径，可能防止了死亡，减少了超额死亡人数。例如，在富裕的城市县域内生活的居民拥有居家办公和避免家庭拥挤的特权，这可能降低他们死于流感的风险。流感每年通常会造成50,000人死亡。
事实上，2020至2021年流感季非常轻微，原因可能是社交隔离。另外一种可能的解释是，在疫情前两年后期，某些地区的死亡人数少于预期，原因是在这些地区健康状况最差的居民已经死于新冠。这些解释表明，即使在新英格兰和大西洋中部各县，虽然官方记录的新冠死亡人数高于预估超额死亡人数，在其他死亡人数减少时，许多新冠死亡人数依旧有可能发生。
明确死亡人数为什么重要
最后，确定有多少人死于新冠疫情，是一项重要的科学事业，会产生重大政治影响。明确新冠死亡人数和死亡地点具有广泛的意义，能够知道目前对疫情响应资源的分配方式，并为未来的公共卫生紧急事件做好准备。
因此，我们认为，科学界应该仔细审查统计新冠死亡人数背后的科学严谨性。由于疫情被严重政治化，在发表新冠死亡人数被高估或低估的言论时，都应该慎之又慎。
最后，我们的团队进行的研究以及我们参与开展的调查报告发现，新冠死亡人数被低估的情况在黑人、拉丁裔和美洲原住民社区以及低收入地区明显更为普遍。主张新冠死亡人数被高估的说法，破坏了为核对这些社区被低估的死亡人数和保证向受影响最严重的地区分配资源所做的努力。例如，如果一个人的死亡证明上未注明死因为新冠，他们的家人就不具备参加疫情社会保障项目的资格，例如美国联邦应急管理署（FEMA）的葬礼援助项目。
为了了解在疫情期间，美国公共卫生制度的成功和不足之处，充分统计新冠导致的死亡人数必不可少。除此之外，死者的亲友和爱人，也应当知道新冠造成的真正损失。
安德鲁·斯托克斯现任波士顿大学全球健康系助理教授；迪耶莱·伦德伯格现任波士顿大学全球健康系助理研究员；伊丽莎白·里格利-菲尔德现任明尼苏达大学社会学助理教授；陈毅恒（音译）现任加州大学旧金山分校流行病学和生物统计学研究数据专家。（财富中文网）
本文根据知识共享许可转载自The Conversation。
翻译：刘进龙
审校：汪皓
自2020年3月新冠疫情被宣布构成全球大流行以来，有一个话题不断引起争论，那就是美国官方公布的新冠死亡人数统计数据，是否准确体现了与引发新冠的SARS-CoV-2病毒有关的死亡人数。
有些政客和许多公共卫生从业者认为，新冠死亡人数被高估。例如，2023年1月，《华盛顿邮报》发表的一篇评论文章称，新冠死亡人数不仅包括死于新冠的人数，还包括死于其他原因但恰好感染新冠的人数。
但大多数科学家认为，新冠死亡人数被低估，因为这些数据没有包含被错误分类为其他死因的新冠死亡人数。
我们是由波士顿大学（Boston University）、明尼苏达大学（University of Minnesota）、加州大学旧金山分校（University of California San Francisco）等机构组成的研究团队的成员，自疫情爆发以来持续跟踪新冠死亡人数。团队的主要目标是评估美国新冠死亡人数是否被低估，以及哪些地区的死亡人数被低估。
研究超额死亡人数
研究这个问题的途径之一是分析人口健康研究人员所说的超额死亡率。这个指标将疫情期间的死亡人数与根据疫情之前的趋势预测的死亡人数进行对比。
超额死亡率评估了直接死于新冠的人数，或通过间接途径死亡的人数，例如在新冠高峰期间避免前往医院的患者。确定死因是一个复杂的过程，而记录是否死亡却非常直接。因此，计算超额死亡人数被认为是对疫情死亡人数偏差最小的估算。
一般而言，如果新冠死亡人数超过超额死亡人数，这意味着新冠死亡人数可能被高估，当然有一些重要的说明，会在下文一一解释。如果超额死亡人数超过新冠死亡人数，后者可能被低估。
在最近发布的一项尚未经过同行审议的研究中，我们的团队发现，在2020年3月至2022年2月疫情的前两年，美国超额死亡人数为996,869至1,278,540人。其中866,187人的死亡证明中确认的死因为新冠。这意味着超额死亡人数比新冠死亡人数多130,682至412,353人。疫情第一年和第二年，两者之间存在巨大差距。这意味着即使在经过疫情最初几个月的混乱之后，新冠疫情死亡人数依旧被低估。
多项重要研究也认为，在新冠疫情前两年，全美超额死亡人数超过了新冠死亡人数。我们的团队开展的初步分析发现，超额死亡人数和新冠死亡人数的差距延续到了疫情第三年。这表明，新冠死亡人数现在依旧被低估。
理解数据差异
解释超额死亡人数与公布的新冠死亡人数之间的差异是一项更具有挑战性的任务。但有许多证据证明，数据的差异基本代表未被统计的新冠死亡人数。
我们在最近的一项研究中发现，超额死亡人数在官方公布的新冠死亡人数之前达到最高峰。即使与阿尔兹海默症等死因有关的超额死亡人数也是如此，因为在疫情期间，由于患者避免去医院或其他行为变化，这类超额死亡人数不太可能快速发生变化。
这一研究结果与我们的观察结果一致。我们发现，在新冠高峰开始时，新冠死亡病例可能未被确认或被错误划分为其他死因。目前，医疗提供商和死亡调查员在社区中进行新冠检测的频率可能降低。如果超额死亡的死因并非新冠疫情，则在新冠疫情高峰时期，超额死亡人数应该维持相对稳定，或者在疫情高峰过后，由于医院人满为患和医疗中断导致死亡，从而使超额死亡人数达到最高峰。
疫情期间，与服药过量等外部死因有关的超额死亡人数同样增多。然而，一项初步研究发现，与超额死亡人数的整体增长幅度相比，这类死亡人数的增长幅度相对较小。因此，外因死亡人数也不能解释超额死亡人数和新冠死亡人数之间的差异。
鉴于前文提到的《华盛顿邮报》发表的那篇重要的评论文章，这个证据值得考虑。文章提出，美国的新冠死亡人数被严重高估。作者主张，在某些医院，大范围新冠检测导致感染新冠但死于其他病因的患者，在死亡证明中依旧将新冠作为死因之一。然而，这是一种根本性的误解，将院内死亡人数泛化到全国。
这种过度泛化存在缺陷的一个原因是，院内死亡不同于院外死亡。在医院以外，通常没有新冠检测，死亡调查人员没有接受充分培训，而且对死者不够了解。事实上，我们的研究表明，在医院外，新冠死亡人数被严重低估。
农村地区的验尸官调查报告也显示，在院外死因确认方面存在巨大差异。如果与验尸官自己的政治理念不一致，或者死者家属希望不要提及新冠，有些验尸官甚至会在死亡记录中说明死者未感染新冠。
过度泛化的另外一个问题是地理差异。初步研究表明，美国绝大多数县的超额死亡人数超过新冠死亡人数。尤其是在南方各县、落基山脉各州和农村地区，超额死亡人数超过新冠死亡人数的比例更高。这表明，在这些地区新冠死亡人数可能被低估。
我们的分析结果，至少在非常有限的程度上，可以支持新冠死亡人数有时候被高估的观点。在新英格兰和大西洋中部各州的多个大中型城区，新冠死亡人数超过了超额死亡人数。但美国大多数地区并没有遵循少数几个县的模式。
在新英格兰和大西洋中部各州，将新冠划定为死因的部分死亡人数可能实际上并非死于新冠，对于这种情况可能有其他解释。首先，在这些地区，新冠疫情防控措施，通过与新冠无关的途径，可能防止了死亡，减少了超额死亡人数。例如，在富裕的城市县域内生活的居民拥有居家办公和避免家庭拥挤的特权，这可能降低他们死于流感的风险。流感每年通常会造成50,000人死亡。
事实上，2020至2021年流感季非常轻微，原因可能是社交隔离。另外一种可能的解释是，在疫情前两年后期，某些地区的死亡人数少于预期，原因是在这些地区健康状况最差的居民已经死于新冠。这些解释表明，即使在新英格兰和大西洋中部各县，虽然官方记录的新冠死亡人数高于预估超额死亡人数，在其他死亡人数减少时，许多新冠死亡人数依旧有可能发生。
明确死亡人数为什么重要
最后，确定有多少人死于新冠疫情，是一项重要的科学事业，会产生重大政治影响。明确新冠死亡人数和死亡地点具有广泛的意义，能够知道目前对疫情响应资源的分配方式，并为未来的公共卫生紧急事件做好准备。
因此，我们认为，科学界应该仔细审查统计新冠死亡人数背后的科学严谨性。由于疫情被严重政治化，在发表新冠死亡人数被高估或低估的言论时，都应该慎之又慎。
最后，我们的团队进行的研究以及我们参与开展的调查报告发现，新冠死亡人数被低估的情况在黑人、拉丁裔和美洲原住民社区以及低收入地区明显更为普遍。主张新冠死亡人数被高估的说法，破坏了为核对这些社区被低估的死亡人数和保证向受影响最严重的地区分配资源所做的努力。例如，如果一个人的死亡证明上未注明死因为新冠，他们的家人就不具备参加疫情社会保障项目的资格，例如美国联邦应急管理署（FEMA）的葬礼援助项目。
为了了解在疫情期间，美国公共卫生制度的成功和不足之处，充分统计新冠导致的死亡人数必不可少。除此之外，死者的亲友和爱人，也应当知道新冠造成的真正损失。
安德鲁·斯托克斯现任波士顿大学全球健康系助理教授；迪耶莱·伦德伯格现任波士顿大学全球健康系助理研究员；伊丽莎白·里格利-菲尔德现任明尼苏达大学社会学助理教授；陈毅恒（音译）现任加州大学旧金山分校流行病学和生物统计学研究数据专家。（财富中文网）
本文根据知识共享许可转载自The Conversation。
翻译：刘进龙
审校：汪皓
Since the COVID-19 pandemic was declared in March 2020, a recurring topic of debate has been whether official COVID-19 death statistics in the U.S. accurately capture the fatalities associated with SARS-CoV-2, the virus that causes COVID-19.
Some politicians and a few public health practitioners have argued that COVID-19 deaths are overcounted. For instance, a January 2023 opinion piece in The Washington Post claims that COVID-19 death tallies include not only those who died from COVID-19 but those who died from other causes but happened to have COVID-19.
Most scientists, however, have suggested that COVID-19 death tallies represent underestimates because they fail to capture COVID-19 deaths that were misclassified to other causes of death.
We are part of a team of researchers at Boston University, University of Minnesota, University of California San Francisco and other institutions who have been tracking COVID-19 deaths since the beginning of the pandemic. A major goal for our team has been to assess whether the undercounting of COVID-19 deaths has occurred, and if so in which parts of the country.
Examining excess deaths
One way to examine the issue is to look at what population health researchers call excess mortality. It’s a measure which, in this case, compares the number of deaths that occurred during the pandemic to the number of deaths that would have been expected based on pre-pandemic trends.
Excess mortality captures deaths that arose from COVID-19 directly or through indirect pathways such as patients avoiding hospitals during COVID-19 surges. While determining a cause of death can be a complex process, recording whether or not someone died is more straightforward. For this reason, calculations of excess deaths are viewed as the least biased estimate of the pandemic’s death toll.
As a general rule of thumb – with some important caveats that we explain below – if there are more COVID-19 deaths than excess deaths, COVID-19 deaths were likely overestimated. If there are more excess deaths than COVID-19 deaths, COVID-19 deaths were likely underestimated.
In a newly released study that has not yet been peer-reviewed, our team found that during the first two years of the pandemic – from March 2020 to February 2022 – there were between 996,869 and 1,278,540 excess deaths in the U.S. Among these, 866,187 were recognized as COVID-19 on death certificates. This means that there were between 130,682 and 412,353 more excess deaths than COVID-19 deaths. The gap between excess deaths and COVID-19 deaths was large in both the first and second years of the pandemic. This suggests that COVID-19 deaths were undercounted even after the pandemic’s chaotic early months.
Major studies have also concluded that excess deaths exceeded COVID-19 deaths at the national level during the first two years of the pandemic. And preliminary analyses by our team have found that the gap between excess deaths and COVID-19 deaths has persisted into the third year of the pandemic. This suggests that COVID-19 deaths are still being undercounted.
Making sense of the discrepancy
Explaining the discrepancy between excess deaths and reported COVID-19 deaths is a more challenging task. But several threads of evidence support the idea that the difference largely reflects uncounted COVID-19 deaths.
In a recent study, we found that excess deaths peaked immediately before spikes in reported COVID-19 deaths. This was the case even for excess deaths associated with causes like Alzheimer’s disease that are unlikely to rapidly change due to patients avoiding hospitals or other changes in behavior during the pandemic.
This finding aligns with the observation that COVID-19 deaths may go unrecognized – and be misclassified to other causes of death – at the beginning of COVID-19 surges. At this time, COVID-19 testing may be less frequent in the community, among medical providers and among death investigators. If excess deaths were not caused by COVID-19, they would instead either remain relatively constant during COVID-19 surges or they would peak afterwards when hospitals were overcrowded and deaths may have resulted from health care interruptions.
Excess deaths related to external causes of death such as drug overdose also increased during the pandemic. However, a preliminary study found that the scale of this increase was small relative to the overall increase in excess deaths. So deaths from external factors alone cannot explain the gap between excess and COVID-19 deaths.
This evidence is worth considering in light of the prominent opinion piece in the Washington Post mentioned earlier, which suggests that the U.S.‘s tally of COVID-19 deaths is a substantial overcount. The author argues that in some hospitals, widespread COVID-19 testing has led patients with COVID-19 who died of other causes to still have COVID-19 included as a cause on their death certificate. There is a fundamental misunderstanding, however, in generalizing these hospital deaths to the entire country.
One reason this overgeneralization is flawed is because hospital deaths are distinct from out-of-hospital deaths. In out-of-hospital settings, COVID-19 testing is often lacking and death investigators have less training and less information about the deceased. In fact, our research suggests that COVID-19 deaths are largely undercounted in out-of-hospital settings.
Investigative reporting among coroners in rural areas has also revealed significant variability in out-of-hospital cause of death assignment. Some coroners have even gone on record to state that they do not include COVID-19 on death records if it contradicts their own political beliefs or if families wish for it to be omitted.
The other problem with the overgeneralization is geographic. Our preliminary study demonstrates that excess deaths exceeded COVID-19 deaths in the vast majority of counties across the U.S. In particular, counties in the South, the Rocky Mountain states and rural areas had many more excess deaths than COVID-19 deaths. This suggests that COVID-19 deaths were likely undercounted in these areas.
The idea that COVID-19 deaths are sometimes overreported is, to a very limited extent, supported by our analyses. A select number of large and medium-sized metro areas in New England and the mid-Atlantic states have had more COVID-19 deaths than excess deaths. But most of the country has not followed the patterns of this small group of counties.
While it is possible that some deaths assigned to COVID-19 in New England and the mid-Atlantic states were not actually caused by COVID-19, other explanations are also possible. First, COVID-19 mitigation efforts could have prevented deaths in these areas via pathways unrelated to COVID-19, reducing excess deaths. For example, some people living in wealthy, urban counties had the privilege to work from home and avoid household crowding, which may have reduced their risk of dying from flu. Flu is typically responsible for as many as 50,000 deaths each year.
In fact, the 2020-2021 flu season was minimal, likely because of social distancing. Another possible explanation is that later in the first two years of the pandemic, there may have also been fewer deaths than expected in some areas because some of the least healthy people in the area had already died of COVID-19. These alternative explanations imply that, even in those New England and mid-Atlantic counties where more COVID-19 deaths were recorded than estimated excess deaths, many COVID-19 deaths may still have occurred even as other kinds of deaths decreased.
Why it matters
Ultimately, figuring out how many people have died as a result of the COVID-19 pandemic is a major scientific undertaking that has significant political importance. Knowing how many people died and where these deaths occurred has widespread implications for informing how current pandemic response resources are allocated and for preparing for future public health emergencies.
As a result, in our view, it is critical that the scientific community carefully reviews the rigor of the science behind the counting of COVID-19 deaths. Given the intense politicization of the pandemic, claims of overcounting or undercounting need to be made cautiously.
Finally, research by our team and investigative reporting conducted in partnership with our team has found that the undercounting of COVID-19 deaths is significantly more common in Black, Hispanic and Native American communities as well as low-income areas. Claims that COVID-19 deaths have been overcounted undermine efforts to reconcile the undercounts in these communities and to ensure resources are being allocated to those most affected. For example, if a person does not have COVID-19 assigned as a cause on their death certificate, their family is ineligible for pandemic social programs such as the FEMA funeral assistance program.
To understand where the U.S. public health system has succeeded and fallen short during the pandemic, a full accounting of deaths caused by COVID-19 is needed. More than that, families, friends and loved ones of those who have died so far also deserve to know the true toll that COVID-19 has taken.
Andrew Stokes is Assistant Professor of Global Health, Boston University; Dielle Lundberg is Research Assistant in the Department of Global Health, Boston University; Elizabeth Wrigley-Field is Assistant Professor of Sociology, University of Minnesota, and Yea-Hung Chen is Research Data Specialist in Epidemiology and Biostatistics, University of California, San Francisco.
This article is republished from The Conversation under a Creative Commons license.
