Crisis Decision-Making at the Speed of COVID-19: Field Report on Issuing the First Regional Shelter-in-Place Orders in the United States

Abstract

Context: 

In March, 2020, the severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2), the causal agent of coronavirus disease 2019 (COVID-19), was spreading in the Bay Area, especially in Santa Clara County, causing increases in cases, hospitalizations, and deaths.

Program: 

The Association of Bay Area Health Officials (ABAHO) represents 13 Bay Area health jurisdictions.

Implementation: 

On March 15, 2020, the local health officers of 7 ABAHO members (counties of Alameda, Contra Costa, Marin, San Francisco, San Mateo, and Santa Clara and the city of Berkeley) decided to issue legal orders on March 16 for 6.7 million residents to shelter in place to prevent the spread of SARS-CoV-2, the causal agent of COVID-19. The Bay Area was the first region in the United States to shelter in place, and within days, other regions in the United States followed.

Evaluation: 

Subsequent comparative analyses have confirmed that acting early in issuing shelter-in-place orders prevented a large number of cases, hospitalizations, and deaths in the Bay Area throughout the United States. The quality of a decision—in this case, for crisis decision making—cannot be judged by the outcome. A good decision can have a bad outcome, and a bad decision can have a good outcome. The quality of a decision depends only on the quality of the decision-making process at the time the decision was made.

Discussion: 

In this Field Report, we review how we made this collective decision. With the benefit of hindsight and reflection, we recount our story through the lens of public health legal authority, meta-leadership, and decision intelligence. Our purpose is to improve the crisis decision-making skills of public health officials by improving how we make high-stakes decisions each day in our continuing fight to contain the SARS-CoV-2 pandemic, to save lives, and to eliminate COVID-19 racial/ethnic inequities.

On March 16, 2020, the health officers of 6 Bay Area counties (Alameda, Contra Costa, Marin, San Francisco, San Mateo, and Santa Clara) and the city of Berkeley issued legal orders, effective March 17, for all residents to shelter in place (SIP) to prevent the spread of SARS coronavirus-2 (SARS-CoV-2), the causal agent of coronavirus disease-19 (COVID-19).¹ The Bay Area was the first region in the United States to SIP and shortly thereafter the state of California and New York City followed with similar orders issued on March 19 and March 20 (effective March 22), respectively.^2,3

Numerous news stories have been published depicting different aspects of this event.^4–8 To summarize, on the morning of Sunday, March 15, Dr Grant Colfax, San Francisco Director of Health, requested that Dr Tomás Aragón, San Francisco Health Officer, convene a call with the county health officers from Santa Clara (Dr Sara Cody) and San Mateo (Dr Scott Morrow) to discuss how to improve coordination of health officer orders restricting mass and social gatherings. San Mateo had just restricted gatherings to 50 persons, the lowest in the Bay Area at that time. We briefly discussed SIP and universal masking as 2 community mitigation measures that we had not yet implemented. SIP was a natural extension of restricting gatherings, and we were considering this as a regional recommendation.

A few hours later, Dr Cody convened an emergency call with the health officers from 7 Bay Area health jurisdictions urging that the SIP must not only be a legal order but also be issued immediately. In the Bay Area, Santa Clara County had recorded the first case of community transmission and the first death and was experiencing the largest number of cases, hospitalizations, and deaths. Based on the explosive surge of hospitalizations and deaths in Italy, and modeling calculations of doubling times measured in days, she made the convincing case that “hours matter” and that we must mobilize a multijurisdictional legal team to prepare and issue orders by the next day at the latest. After some deliberation, we agreed. The effort to draft orders was led by the Santa Clara Office of the County Counsel. On March 16, 2020, simultaneous press conferences were held by the 7 Bay Area health officers at the Santa Clara County Sheriff's Department¹ and by Mayor London Breed and Dr Grant Colfax at San Francisco's City Hall.⁹ The SIP orders were the culmination of a series of difficult decisions to issue community mitigation orders.

We decided to issue SIP orders at a time when there were relatively few cases in the Bay Area (Figure 1). Subsequent comparative analyses have confirmed that acting early in issuing SIP orders prevented a large number of cases, hospitalizations, and deaths.^10–12 The hospital and health care systems in the Bay Area were never strained in terms of the availability of staffed hospital and intensive care unit beds, as well as supply of mechanical ventilators. Compared with other urban centers during the same period, COVID-19 case and hospital mortality rates were lower in the Bay Area.¹³

FIGURE 1: The Flattening of the Bay Area Epidemic Curve of SARS-CoV-2 Cases (Left) and Deaths (Right)^{a a}The shelter-in-place order was issued on March 16, 2020. In mid-June, the Bay Area and California experienced a surge of cases and deaths due to increases in social gatherings, increases in economic activity from business sector reopenings, and community spread especially among Latinx households and essential workers. From San Francisco Chronicle Coronavirus Tracker, September 6, 2020. Available from: https://projects.sfchronicle.com/2020/coronavirus-map. Used with permission.

Although the Bay Area has been lauded for acting early and decisively, at the time we made our decision, there was tremendous uncertainty whether our actions would avert a regional public health catastrophe. For example, New York City's SIP started on March 22—5 days after the Bay Area's—yet, they experienced significantly higher per capita case and death rates than San Francisco, which, after New York City, is the second densest city in the United States. Table 1 summarizes the COVID-19 experience comparing San Francisco with New York City through May 31, 2020.¹² A key point is that the quality of a decision—in this case, for crisis decision making—cannot be judged by the outcome. A good decision can have a bad outcome, and a bad decision can have a good outcome.^14–16 The quality of a decision can only be judged by the quality of the decision-making process at the time the decision was made. Public health officials should know what goes into making good decisions in the face of high stakes, high uncertainty, extreme time constraints, and multiple competing objectives.

TABLE 1 - Comparison of San Francisco and New York City COVID-19 Experience as of May 31, 2020^a

Measure	San Francisco	New York City
COVID-19 tests per 100 000 residents	7 746	11 745
Positive cases per 100 000 residents	291	2 409
Deaths per 100 000 residents	5	258
Population	883 305	8 336 817
Population density (people per square mile)	18 849	27 751
Population aged 65 y or older	16%	14%
Population that takes public transportation to work	34%	56%
Median household income	$112 376	$60 762
Bachelor's degree or higher	60%	37%
Average high/low temperature in March	62°F/49°F	52°F/36°F

^aSan Francisco and New York City detected their first coronavirus cases in early March. Both issued shelter-in-place orders within 5 days of each other. However, the per capita COVID-19 case and death rates were very different. From Balzer and Whitcomb.¹²https://theconversation.com/coronavirus-deaths-in-san-francisco-vs-new-york-what-causes-such-big-differences-in-cities-tolls-138399

The purpose of this Field Report is to reflect on our experience of crisis decision making applied to the SARS-CoV-2 pandemic and to review key elements of good crisis decision making so that we become more effective at protecting and promoting health in the setting of complex public health emergencies. We review context (legal authority), meta-leadership, and decision intelligence.

Crisis Decision Making

Decision making is our most important daily activity. Decisions drive vision, strategy, execution, problem solving, performance, evaluation, and continuous improvement. Decisions can be rapid and guided by experience and intuition, or they can be the product of a slow, deliberative team process. So, what is a decision?

“A decision is a choice between two or more alternatives that involves an irrevocable allocation of resources.”¹⁵ Every decision has a causal assumption, a prediction, and an opportunity cost and involves human psychology, intuition, and emotions. For example, we might choose option A over option B because we predict option A will achieve Y with probability P (often unstated). The opportunity cost is the lost benefit of the better option not chosen or not considered. Implicit cognitive biases, individually and collectively, can drive the strength of our causal beliefs, predictions, and decisions.^17,18 Therefore, to improve team crisis decision making, we also need to strengthen our capabilities in causal reasoning, probabilistic reasoning, balancing trade-offs from competing objectives, and leading team deliberations in the face of volatility, uncertainty, complexity, and ambiguity (so-called “VUCA”¹⁹).

What public health officials bring to the table are the scientific principles of epidemiology and biostatistics that support causal and probabilistic reasoning, respectfully. Causal reasoning appears in questions such as: What are the key drivers? (root causes); Why and how is this happening? (root causes); Asking “why?” 5 times. (“The 5 Whys”); Does this countermeasure work? (efficacy); Will it work in the real world? (effectiveness); and Which intervention is more cost-effective? (efficiency). Probabilistic reasoning appears in questions such as: What are the chances this countermeasure will work? (prediction); What are the sensitivity and specificity of this test or classification? (proportion of false-negative and/or false-positive of any test result or conjecture); and What are the chances my hypothesis (conjecture, suspected diagnosis) is correct, given the evidence? This is called evidential or diagnostic reasoning and is the reverse of causal (predictive) reasoning. Unfortunately, human beings are poor at probabilistic reasoning, especially with evidential reasoning that requires Bayes theorem to calculate probabilities. Humans are unaware of this cognitive limitation; hence, we are overconfident in our assessments and, with confirmation bias, may develop certainty and rigid beliefs.¹⁷ For these reasons, we must approach VUCA with genuine intellectual humility and embrace curiosity over certainty. In fact, in the face of VUCA, key decisions are better managed like hypotheses to be tested.

Because of VUCA, crisis decision making is often more art than science, and the “art” depends on effective interpersonal skills for building consensus and managing conflict. Decisions in organizational environments evolve within a historical, political, and cultural context that may constrain or support good decision making.²⁰ In the sections that follow, we review public health aspects of crisis decision making relevant to the SARS-CoV-2 pandemic, namely, legal authority, meta-leadership, and decision intelligence. Other interpersonal aspects of crisis decision making (eg, moral, ethical, and political) are covered elsewhere.^21,22

Public Health Legal Authority

The Bay Area local health officers (LHOs), under California law, have significant autonomy to impose broad legal mandates to protect the health of the public. In California, there are 61 local health jurisdictions: 58 counties and the cities of Berkeley, Long Beach, and Pasadena (San Francisco is a city and county). Local health authority is vested at the county level, but a city can become a health jurisdiction.

By California law, every health jurisdiction must employ a physician LHO who, upon appointment, is also an agent of the state. In most counties, the LHO is appointed by the county board of supervisors. The California Conference of Local Health Officers (CCLHO) was established by statute in 1947 to advise the Department of Public Health and other officials of federal, state, and local agencies, the legislature, and other organizations on all matters affecting health. LHOs have a long history of working together to pursue common goals at the state and regional levels.

California LHOs are authorized to control the spread of communicable diseases and may “take measures as may be necessary” including isolating cases, quarantining contacts, restricting mass gatherings, commandeering assets, and far-reaching actions such as issuing orders to SIP. This broad authority can be exercised without state prior approval and is independent of local elected officials. Ultimately, the state public health officer has the authority to override the LHO's authority, but this is rarely done.

The public health governance structure in California enabled the LHOs of the Bay Area region to quickly and decisively act to issue SIP orders to protect the public's health and health care system from the type of surge of COVID-19 cases, hospitalizations, and deaths seen in New York City. Similar to CCLHO, the Association of Bay Area Health Officials (ABAHO) has a long history of close collaboration and decisive action that dates back to the HIV/AIDS epidemic in the 1980s.²³ With the SARS-CoV-2 pandemic, ABAHO started meeting twice weekly. This culminated with 7 of 13 ABAHO health jurisdictions to move rapidly to SIP. The 7 counties that bordered the Bay moved together for speed of execution, not because of disagreement. Within days, the state and other counties followed.

Meta-leadership

CCLHO and ABAHO are concrete examples of statewide and regional meta-leadership networks that respond, adapt, and scale to public health threats. The meta-leadership framework was developed from the empirical study of leadership during large-scale disasters (eg, World Trade Center terrorist attack, Hurricane Katrina, Boston Marathon bombing).²² Meta-leadership has 3 dimensions (Figure 2): the person (eg, health officer), the situation (eg, SARS-CoV-2 pandemic), and connectivity (stakeholder relationships). Connectivity has 4 facets: leading down (your staff), leading up (your bosses), leading across (your institutional silos), and leading beyond (cross-sector relationships in the community and beyond).

FIGURE 2: Dimensions of Meta-leadership^{a a}Dimensions of meta-leadership: (1) the person, (2) the situation, and (3) connectivity (leading down, leading up, leading across, and leading beyond). ^aAdapted with permission from Marcus et al.²²

To be effective public health leaders, LHOs must be effective meta-leaders. For years, Bay Area health officers have been preparing for pandemic influenza.²⁴ This transitioned into ABAHO subcommittees for public health preparedness, mass prophylaxis, and public information. Over the years, the subcommittees used meta-leadership skills to develop a governance structure and a strategic plan and obtain grant funding for regional coordination and created a Multi-Agency Coordination (MAC) group that conducted public health emergency exercises to test our regional readiness. After several years of external grant funding, agreement was made for several counties to contribute funds to hire a regional preparedness coordinator. In January 2020, ABAHO activated the MAC group to meet regularly to discuss emerging issues, solve problems, and share best practices. In the SARS-CoV-2 pandemic, the Bay Area LHOs' strong, preexisting relationships and collaborations enabled the spontaneous activation of what Dr Leonard Marcus calls “swarm leadership”^22,25:

1. Unity of mission (focused on saving lives);
2. Generosity of spirit and action;
3. Staying in their own lanes;
4. No ego, no blame; and
5. Trust-based relationships.

Swarm leadership has been observed in good disaster responses. The concept is based on the observation in nature of “swarm intelligence”: “how creatures intuitively accomplish remarkable tasks when no one of them is in charge.”²² “Staying in their own lane” means we stayed focused on public health principles and legal authority, and we respected, valued, and leveraged the expertise and authority of partner agencies. Our trust-based relationships were forged over the years by informal dinner meetings held every other month, as well as health officer training that included sessions on humility, cultural humility,²⁶ cognitive biases,¹⁷ public health emergency readiness, and performance improvement.²⁷

On March 15, we faced high-stakes VUCA. Dr Cody stated, “What we're doing is not working.” Without knowing it at the time, we managed uncertainty by deliberation and “driving to the knowns.” Figure 3 summarizes 2 dimensions of uncertainty: (1) What can be known? and (2) What is known? This creates 4 categories of exploration and learning. Therefore, to navigate uncertainty, ask these questions to identify the data and information that we should know:

1. The known knowns: What do we know that we know? Use it.
2. The known unknowns: What do we know that we don't know? Assemble it (ie, the information is available to us).
3. The unknown knowns: What do we not know that others know? Seek it (eg, information that is hidden from us, including blind spots).
4. The unknown unknowns: What do we not know that others don't know either? Imagine it (eg, “what if” scenario planning).

FIGURE 3: “Driving to the Knowns” Is a Framework for Managing Uncertainty in Information or knowledge^{a a}“Driving to the knowns” framework: (1) known knowns: What do we know that we know? Use it. (2) known unknowns: What do we know that we don't know? Assemble it. (3) unknown knowns: What do we not know that others know? Seek it. (4) unknown unknowns: What do we not know that others don't know either? Imagine it. Adapted with permission from Marcus et al.²²

For 1 (known knowns): COVID-19 was overwhelming Italy with hospitalizations and deaths.²⁸ SIP worked in China and in the 1918 influenza pandemic.²⁹ In San Francisco, on March 10, 2020, Tomas Pueyo posted an article on Medium³⁰ and within 9 days it was read 30 million times.³¹ His article raised awareness of SARS-CoV-2's explosive, exponential growth and the implications of waiting even a few days to act. For 2 (known unknowns): we did not have sufficient testing to be able to identify who was infectious and who was not. For a case in a school, we did not know whether this was an introduction or the “tip of an iceberg.” For 3 (unknown knowns): we sought input and feedback from key stakeholders including attorneys, the state health department, and county administrators. Some stakeholders advocated waiting a few days to improve preparation and execution. For 4 (unknown unknowns): we imagined different scenarios; for example, the downside of waiting a few days. We decided not to wait and the orders were issued the next day on March 16, 2020.

Our greatest vulnerabilities are the unknown knowns and the unknown unknowns because we are completely unaware of critical gaps in our knowledge. Address these gaps by convening teams that have cognitive, technical, and cultural diversity. Ensuring psychological safety,³² and modeling intellectual humility,³³ builds trust and enables vigorous deliberation, constructive conflict, and comprehensive consensus (shared understanding and commitment).²⁰

Decision Intelligence

For public health officials, decision intelligence is the practical integration of problem solving, decision quality, and performance improvement. Many public health officials use components of decision intelligence but are unaware of its integrated structure. Figure 5 graphically depicts the problem-solving structure as a causal graph that guides causal and diagnostic reasoning. For example, in February 2020, we were screening passengers arriving to the United State from Wuhan, China. If a passenger had a fever and cough (problem), we would interview the passenger to gather information and decide whether to classify them as a “person under investigation” (updated problem definition). The potential consequences were prioritized (eg, infected, infectiousness, medical complications). We conducted a root cause analysis to assess their exposures, prioritized causal hypotheses (differential diagnosis), and selected diagnostic tests (eg, nucleic acid test). Based on what we decided (hypothesized) were the problem, consequences, and root causes, we selected, implemented, and evaluated countermeasures appropriate for each (eg, quarantine and isolation). Clinical providers also use decision intelligence for diagnostic evaluations.

Similarly, complex problem-solving is a series of important, causally linked decisions to (a) select and focus on the right problem, consequences, and root causes, and (b) design, evaluate, and improve countermeasures (prevention, control, mitigation) to achieve the primary objectives while minimizing harmful outcomes and unintended consequences. Decision quality is understanding and improving the requirements of making good decisions (Figure 4) even when the decisions are intuitive and fast. At a minimum, a decision quality checklist (Table 4) improves the quality of decisions at any stage of problem solving.

FIGURE 4: Decision Quality Requirements^{a a}Decision quality requirements: (1) frame the decision problem or opportunity, including identifying values and setting objectives; (2) gather relevant data and information; (3) generate creative, doable alternatives (choices); (4) conduct sound reasoning to select or prioritize the best alternatives to achieve the objectives; (5) involve the right stakeholders and build consensus (commitment to action); and (6) understand the consequences and trade-offs (prospects).

FIGURE 5: Causal Graph for Problem Solving^{a a}Problem solving is a series of causally linked decisions to (a) select and focus on the right problem, consequences, and root causes, and (b) design, evaluate, and improve countermeasures (prevention, control, mitigation) to achieve the primary objectives while minimizing harmful outcomes. Countermeasures can also be described as primary, secondary, and tertiary prevention.

Team problem solving is a series of deliberative decision processes that are divergent (generating creative ideas and options) and convergent (selecting or prioritizing options for action). First, selecting the right problem to tackle is a decision problem, and it is the most important. For example, if the primary problem had been defined as hospitalizations, and deaths as the consequences, then we might have limited our countermeasures to increasing hospital bed and intensive care unit surge capacity as the primary strategy to save lives. In contrast, with a public health prevention mindset, we defined the problem as uncontrolled community transmission of SARS-CoV-2, with the consequences being the number of cases, hospitalizations, and deaths. Second, a root cause analysis identifies what causes the problem. Third, selecting countermeasures for root causes, problem, and consequences is, again, a decision problem. The key point is that crisis problem-solving is a series of linked, interdependent decisions, each with quality requirements. The quality of a decision can only be as strong as its weakest link, and the quality of problem solving can only be as strong as the quality of those decisions.

Put together, decision intelligence is the integration of problem solving and decision quality within a performance improvement framework, ensuring quality and continuous improvement in crisis decision making. Table 2 summarizes decision intelligence as Plan-Do-Study-Act (PDSA) problem solving, and Table 3 summarizes the status of our general decision intelligence on the morning of March 15, 2020. The main problem was, as stated by Dr Cody, “What we're doing is not working” (ie, uncontrolled community transmission of SARS-CoV-2 with exponential growth). Santa Clara was experiencing a surge of hospitalizations and deaths (consequences), and in 1 or 2 weeks the other Bay Area counties “would be Santa Clara” (prediction). The experiences of China and Italy provided a preview of the expected dire consequences (prediction). The Bay Area hospital systems were not sufficiently prepared to mitigate large surges of critical patients. Countermeasures to control community transmission were not in place, including widespread testing, ample number of trained case investigators and contact tracers, and resources to help people isolate, quarantine, and SIP. At the time, our understanding of transmission mechanisms as causal drivers was limited (asymptomatic and presymptomatic infectiousness, virus aerosolization). The efficacy of face masks, physical distancing, and increased ventilation as prevention countermeasures had not been established yet. In short, there were too many unknowns, high-stakes consequences, and no time for further deliberation.

TABLE 2 - Decision Intelligence Is the Integration of Problem Solving and Decision Quality Within a Performance Improvement Framework

PDSA	Components of Problem Solving	Key decision questions
Plan	Problem definition	What is main problem to solve?
	Consequence (risk) analysis	What are the consequences?
	Root cause (diagnostic) analysis	What causes the problem?
	Countermeasure design and testing	What strategies and actions work?
Do	Countermeasure implementation	How do we deploy countermeasures?
Study	Countermeasure (causal) evaluation	How do we measure effectiveness?
Act	Act on what you learn to improve	How do we learn and improve?

TABLE 3 - Decision Intelligence Problem Solving Applied to the SARS-CoV-2 Pandemic in Bay Area on March 15, 2020

Causal Components	Description	Countermeasure	Decision Status
Problem	Uncontrolled transmission	Infection control in nursing homes	Y
		Community and health care masking	N
Causes	Mobility outside the home	Shelter in place (key decision problem)	TBD
	Social and mass gatherings	Numerical restrictions	Y
		School closures	Y
	Personal behaviors	Hand washing and hygiene	Y
		Respiratory hygiene and cough etiquette	Y
	Contaminated fomites	Environmental disinfection	Y
Consequences	Cases	Testing and case finding	Y
		Case investigation and isolation	Y
		Contact tracing and quarantine	Y
	Hospitalizations	Supportive treatment	Y
	Deaths	Intensive care unit	Y
		Mechanical ventilation	Y

Abbreviation: TBD, to be determined.

TABLE 4 - Decision Quality Requirements for the Option to Issue an SIP Order^a

Requirement	Question and description	Example
1. Frame	What are we deciding and why? (values, perspective, scope, objectives)	SIP options; save lives
2. Information	What do we need to know? (data, information, knowledge, and wisdom)	Known knowns, known unknowns, unknown knowns, unknown unknowns
3. Alternatives	What choices do we have? (creative, doable alternatives)	No SIP, SIP-Rec, SIP Order-Now, SIP Order-Wait
4. Reasoning	Are we thinking straight? (sound reasoning: evaluate against objectives, weigh trade-offs)	See “Consequence”
5. Commitment	Is there commitment to action? (consensus; involves right people to decide and execute)	Do we have consensus? Do we have political and public support?
6. Prospects	What future states do we care about? (possible future states, results, impacts)	What are our goals? What the opportunity costs? What are the unintended consequences?

Abbreviations: Rec, recommendation; SIP, shelter in place.

^aAlso see Figure 4.

At the same time, we also recognized that SIP was a drastic measure with significant collateral social harm. If implemented right away, the benefits of interrupting transmission would be maximized. We agreed that if implemented with delay, the benefits drop significantly, but our communities would suffer the same social harms. Although we did not use this practical tool at the time, a Consequence Table (CT) (Table 5) could have been used to evaluate alternatives and the trade-offs.¹⁶ The CT has 3 sections: objectives (value-based goals), alternatives (options), and consequences (outcomes). The outcomes can be quantitative or qualitative. Given the time urgency and limited state of knowledge in the face of VUCA, we relied on our collective knowledge and intuition. At worst, a CT organizes your team reasoning for deliberation and selection, while reducing groupthink.¹⁸ Sound reasoning is a decision quality requirement, and a CT is a field-tested tool for this purpose. To evaluate different combinations of alternatives (“policy options”) consider using a Strategy Table (Table 6 in Supplemental Digital Content, available at https://links.lww.com/JPHMP/A747).

TABLE 5 - Consequence Table Example Applied to the SIP Decision^a

Objectives (Value-Based Goals)	Alternatives
	No SIP	SIP Rec-Now	SIP Order-Now	SIP Order-Wait
Minimize cases and deaths	−2	0	+2	−1
Minimize hospital/ICU surge	−2	0	+2	−1
Minimize economic disruption now	+2	0	−2	−2
Minimize economic disruption later	−2	0	+2	+2
Total	−4	0	+4	−2

Abbreviations: ICU, intensive care unit; Rec, recommendation; SIP, shelter in place.

^aExample applied to the SIP decision: objectives (column 1), alternatives (columns 2-5), and consequences (outcomes as qualitative assessment of the extent that an alternative achieves each objective: +2 = definitely yes; +1 = likely yes; 0 = neutral or not sure; −1 = likely no; and –2 = definitively no).

As a team, we decided to issue SIP orders within 24 hours. Through the SIP orders, we achieved the following objectives: (1) flattened the curve of cases, hospitalizations and deaths in the Bay Area; (2) provided time for hospital systems to prepare for future surges; (3) provided time to learn about SARS-CoV-2 biology and transmission and to implement new countermeasures such as universal masking; (4) provided time to build capacity for widespread testing, case investigation, contact tracing, and support for isolation and quarantine; and (5) provided time for the development of pharmaceutical interventions. Today, we are now able to manage predicted surges as we reopen the economy and schools, and without the need to reinstitute comprehensive SIP restrictions and business closures.

Decision intelligence also requires addressing the psychology, politics, and public health ethics of team decision making. The prerequisites for deliberating difficult decisions is building trust and ensuring psychological safety so that team members can deliberate with candor, generate creative ideas, mitigate groupthink and cognitive biases, and converge toward group consensus.^{14,18,20,32,33} Roberto²⁰ reminds us that a “decisive” leader does not make all the decisions and does not have all the answers. Instead, a decisive leader ensures good decision making processes (ie, decision quality, learning from failures, and continuous improvement).

Marcus et al²² promote a crisis decision-making framework called the POP-DOC loop. The “Thinking steps” are Perceive, Orient, and Predict, and the “Action steps” are Decide, Operationalize, and Communicate. POP-DOC loop is based on the well-known Observe-Orient-Decide-Act (OODA) loop from military strategist Colonel John Boyd.³⁴ The key lesson from these frameworks is the bias to take action, see what happens, learn, and repeat the cycle. The bias-to-action and action-learning mindset is critical to avoid “analysis paralysis.”

Crisis decision making occurs in complex political and organizational environments. One simple tool to support deliberations for multidimensional problems is the (HELP)² checklist (see Table 7 in Supplemental Digital Content, available at https://links.lww.com/JPHMP/A747).²⁷ This ensures that health, equity, ethics, efficiency, legal risks, logistics, political support, and public trust are addressed systematically.

Finally, public health leadership is the “the practice of mobilizing people, organizations, and communities to effectively tackle tough public health challen-ges.”³⁵ Embodying this definition, Dr Cody played a central role as a public health meta-leader to mobilize the Bay Area LHOs to act collectively and decisively to issue the first SIP orders for a region in United States. She facilitated team deliberations consistent with the principles and practices of meta-leadership and decision intelligence and continues to meta-lead at a national level as chair of the Big Cities Health Coalition.¹⁰

Implications for Policy & Practice

• Understand the scope of your legal authority and develop strong relationships with legal counsel.
• Strengthen your meta-leadership network of trust-based relationships.
• Use causal thinking and simple causal graphs to organize your problem solving.
• Decision problems are embedded in the problem-solving process (see Figure 6 in Supplemental Digital Content, available at https://links.lww.com/JPHMP/A747). Sometimes decisions are intuitive and fast (eg, POP-DOC loop); sometimes they require deliberation and decision quality.
• Causal assumptions and predictions are implicit in every decision: make them explicit. Predictions improve learning.
• Human intuition cannot calculate probabilities, especially for novel events (eg, SARS-CoV-2 pandemic); therefore, embrace intellectual humility and choose curiosity over certainty.
• Surround yourself with persons who think differently than you and who can vigorously challenge your thinking and assumptions. Ensure psychological safety.
• Crisis decisions have multiple dimensions and stakeholders that require our attention (see Table 7 in the Supplemental Digital Content, available at https://links.lww.com/JPHMP/A747).

Discussion and Conclusion

On Sunday, March 15, 2020, 7 Bay Area health officers decided to issue the first regional SIP orders in the United States. Shortly afterward, California State and New York City followed. With the benefits of hindsight, we have retold our crisis decision making story through the lens of public health legal authority, meta-leadership, and decision intelligence. California state legal authority gave us local autonomy to act quickly to protect the public from a novel health threat. Our established history of meta-leadership and connectivity enabled the activation of “swarm leadership” on March 15. We came to realize that what we did—to be the first region to issue SIP in the United States—was historic; however, we did not realize that the collective leadership phenomenon even had a name. Finally, in public health, decision intelligence is the practical integration of problem solving and decision quality within a continuous improvement framework (eg, PDSA).

Despite the universal importance of decision making, few of us are trained in decision intelligence. Public health staff (and elected officials) continue to struggle to make better decisions in the face of the SARS-CoV-2 pandemic and VUCA. Many of our current decisions are now also focused on reopening the economy and schools, while trying to sustain and scale our prevention, containment, and mitigation efforts. Therefore, all the meta-leadership skills continue to apply. Decision intelligence provides practical tools to improve crisis decision making. We hope that with experimentation and practice, public health officials will improve their routine team decisions and thereby be capable and ready to make better decisions in new crisis situations.