COVID-19: Org Structure vs. The Mission

Over the last few weeks, I’ve been thinking a lot about the organization chart for public health here in the United States. As I’ve learned more about it, I’ve realized the org structure is completely mismatched with the needs of a pandemic.

At the start of the pandemic, I sort of assumed that the Centers for Disease Control (CDC) was in charge of public health in the United States. I imagined that state and local public health leaders reported to the CDC.

It turns out that my assumptions were completely wrong.

Public health in the United States has a decentralized organizational structure.

In fact, it’s principally managed at the county level (a county is a geographic subdivision of a state and may contain one or more towns or cities).

There are 3,100 counties (or county equivalents) in the U.S. Each county has a public health department or public health district office. This department doesn’t report to the CDC. It reports to the county commission, county mayor, or county executive.

This organization chart works pretty well 99.9% of the time. The problem is that COVID-19 is the 0.1% of the time.

Let me explain.

As the U.S. economy reopens, one of the critical activities needed to manage the virus spread will be contact tracing. When a patient tests positive for COVID-19, someone needs to trace very location the patient has been in the last two weeks and identify every person the patient had contact with.

These people would then be contacted, notified that they may have been exposed to COVID-19, and asked to self-quarantine for two weeks.

The responsibility falls within each county’s public health department… all 3,100 of them.

Each public health department’s objective is to reduce COVID-19 cases in their county.

Let’s walk through how this can be problematic.

I live eight miles from downtown Seattle. I happen to cross a county border between my home and the Seattle airport.

Let’s say that I take a trip to a board meeting just outside of Washington D.C. On the way there, I have a layover in Atlanta where I eat lunch at an airport restaurant. On my way back, I have a layover in Chicago and have dinner at the airport.

Now let’s say that after I get back from my trip, I feel symptomatic. I get a COVID-19 test and the test shows positive.

In my two-day trip, I’ve entered and exited five counties: two in the Seattle metro area, one in Atlanta, one in Washington D.C., and one in Chicago. In addition, I flew on four different airplanes during that time.

Now, a record of my testing positive will go to my county’s public health department. What will they do next?

Remember, their mission is to reduce COVID-19 transmission within the county.

During this trip, I didn’t really interact with anyone in my county.

My county’s public health department lacks the mandate, manpower, and possibly the legal jurisdiction to contact people I’ve had contact with in Atlanta, Washington D.C., and Chicago. At best, they will send a message to the public health departments in those jurisdictions.

Maybe the message will be delivered by fax or email. Maybe it doesn’t get delivered at all.

Question: At this point, who is responsible for making sure that all possible COVID-19 transmission chains that originate with me have been contained?

Answer: Nobody.

Nobody can say, “I am responsible for the Victor Cheng case. He has been in close contact with 60 people. We’ve contacted 49 of them to date. 45 are now in quarantine. We hope to contact the remainder by the end of the day.”

This is a major structural issue with the U.S. public health system organization design.

We have a decentralized system to address a virus that does not respect jurisdictional boundaries.

It’s unlikely that a centuries-old history of decentralized public health in the United States will ever change.

Every org structure has its advantages and disadvantages. The key to designing (or redesigning) an org chart is to understand your objectives.

If you want distributed, local decision-making, you use a decentralized model. If you want consistency and coordination across every region, a centralized model makes more sense.

Every org structure will have its shortfalls. The questions to ask are, which problem do you want the org structure to solve, and which problems do you want to compensate for given the weaknesses of the org structure you choose?

In the case of U.S. public health, a decentralized org structure can still be effective but only by recognizing the weakness of the structure and finding some way to compensate for it.

For example, 3,100 county public health departments could, in theory, all use the same infectious disease contact tracing software system. One county can initiate a record and, using workflow rules, route parts of the case file to other jurisdictions digitally. This could address the coordination and communication issue for a multi-jurisdictional contact tracing case.

Another issue with a decentralized organizational structure is staffing.

99.9% of the time, the staffing levels at county public health departments are adequate to address the needs of that county. However, with pandemics, there can be enormous surges in cases and the need for contact tracing manpower.

For example, the city of Wuhan, China has a population of 11 million people (about three million more than that of New York City). Within weeks of the initial outbreak, the Chinese government deployed 9,000 workers to do contact tracing. That’s in one city.

It would be economically unfeasible to have that kind of full-time staffing in county budgets. 99.9% of the time, those individuals would have nothing to do, and 0.1% of the time, they would be overwhelmed.

Rather than having 3,100 separate public health departments all create their own physical or virtual outbound call centers for contact tracing, it would be far more efficient to centralize, or at least regionalize, those resources.

For example, the state of Massachusetts currently has 1,000 job openings for contact tracers for all counties within the state.

The CDC has staffed a 600-person contact-tracing workforce to do contact tracing for “hot spot” regions that lack the staffing to deal with a sudden surge in cases.

It would make sense to have a much larger centralized (or regionalized) contact-tracing workforce that could be allocated to whichever counties need it the most. The costs for the work could even be billed back to each county on an as-needed basis.

When I work with clients, the issue of optimal organizational structure comes up quite frequently.

As I mentioned earlier, every org structure has its benefits and drawbacks. There is no single org structure that “always” works. The key is to recognize the limitations of the org structure you’ve decided to use and find some other ways to address its limitations.

Let me give you an example.

I was speaking to the CEO of a SaaS (software as a service) company the other day. This company has an organizational structure based around functional areas. One executive runs the sales function. A different executive runs the engineering team. Another runs the customer service organization.

This is a very common org structure for SaaS companies. The CEO was expressing frustration about a particular problem that involved multiple functional areas. The responsibility for the problem fell partly on sales, partly on engineering, and partly on customer service.

So I asked the CEO a simple question, “Could you give me the name of the one person in the company whose job it is to address this problem?”

Silence.

This was a trick question because I already knew the answer. I wanted the CEO to see the problem as I see it. (Yes, I’m sometimes cruel that way with my clients.)

The weakness of a functional org structure is that it deals with cross-functional processes very poorly without some kind of intervention. You can address this weakness in a few different ways:

1) Cross-Functional Projects

Assemble a team with representatives from every functional area, appoint a mid-level rising star to be the day-to-day team manager (this is a great professional development opportunity), and assign a senior executive to represent this project amongst the executive team.

Large Fortune 500 companies have so many cross-functional, cross-regional, and cross-divisional teams that they often have a “project management office.” The project management office addresses the issues that cut across functional boundaries.

2) Cross-Train Your Staff

I like the idea of assigning engineers to do technical support and system installation. It’s quite humbling to take the phone call where the customer yells at you for the bugs in your code. Financial analysts should go on sales calls. A sales manager should sit in on a few product development team meetings.

3) Cross-Locate Your Staff

In many companies, everyone in sales sits together in the office. Everyone in engineering does the same. Another option is to randomly intermix people from different departments throughout the office. Your staff will learn a lot this way.

At my first job after McKinsey, I shared an office with the CFO and COO. At the next company, I shared a pod with a customer success manager and product marketing veteran. On another occasion, I was desk neighbors with the SVP of sales on one side and the VP of channel sales on the other side.

Looking back, a lot of my own professional development came from overhearing the phone calls of my neighbors and asking them questions about it when we were both working late.

This is how I learned about strategies for designing salesforce territories, how to build an indirect sales channel, and how you handle channel conflict. These were all late-night conversations with people who had worked in these areas for three decades.

All of this was entirely a result of where my desk happened to be located.

Every org structure has a sweet spot for what kinds of problems it’s best suited for solving. Every org structure has weaknesses. The key insight in all of this is to be aware of the weaknesses of the org structure you’ve chosen and to find ways to compensate for them.