On delays in access to care in American hospitals

by Hari Balasubramanian

Picture1In my first piece for 3QD, I discussed some aspects of the science of queueing. In this essay, I'd like to ground that discussion in the context of delays that patients routinely experience in American hospitals. In 2010, there were around 130 million visits to the emergency departments (EDs) of hospitals around the country. Nearly 23% of these patients waited an hour or more to see a care provider [link]. Many urban hospitals and individual patients do much worse.

The documentary The Waiting Room foregrounds the human stories that underlie these statistical estimates, and lets patients, their families, nurses, doctors and social workers speak for themselves. The film is shot during a 24-hour period at the emergency department of a safety-net hospital: Highland Hospital in Oakland, California. 241 patients come to seek care during this one day. Most of the time the camera is in the waiting room, where patients reconcile themselves to a long wait; this applies even to a man with a gunshot wound, whose body is turning numb – hard to believe but true.

The problem faced by Highland Hospital is by no means unique. I've heard it many times from hospital administrators and clinicians. My own research on reducing delays in healthcare led me to work with a large hospital which sees nearly 300 patients daily — more than Highland — in its emergency department. I was quite familiar with what I saw in the documentary: the look and feel of the waiting area; the small rooms inside the main care section with beds and equipment, curtained off to provide patients and their families some privacy; the additional hallway beds with no privacy, but nevertheless necessary due to the sheer volume of patients in acute condition; the constant buzz of pagers, movement of personnel, calls for lab analyses and diagnostic scans; the flicker of computer screens with way too much information; the difficulties in deciding whether the patient should be admitted to an inpatient unit or discharged; and if discharged where the patient should go, for some psychiatric or substance abuse patients have no home to return to.


Under federal law, emergency departments cannot deny any patient, irrespective of their ability to pay, and irrespective of the nature of the patient complaint. Treatment is provided first, billing and payment discussed later. So EDs are a safety net for many uninsured or underinsured patients. Even patients who have insurance will use the ED if they cannot access a regular doctor, especially during evening hours or weekends. Emergency departments are supposed to be for very serious and critical cases, but the reality is that they see patients with a lot of other conditions — ranging from a simple sore throat or sprained ankle to more complex chronic cases that may not be critical enough to be termed emergencies but still need attention.

The typical flow in an American ED is somewhat like this. No appointments can be scheduled; a patient simply walks in and registers herself at the front desk. She is then assessed by a nurse, who decides the Emergency Severity Index (ESI) of the patient [link]; this process is called triage. An ESI level 1 patient is “most urgent” requiring immediate attention, while a level 5 patient is the least urgent.

The patient then waits until she is called to the main care area, placed in a bed and seen by a nurse or doctor. Lab and diagnostic tests may be ordered. The patient waits in the bed until the tests are done and assessed. Not surprisingly, ESI level 1, 2 and 3 patients stay much longer and have more diagnostic work done on them. In most cases, the doctor or nurse sees the patient for a few minutes each time, then leaves to see other patients; the care process is thus fragmented as the staff splits their time across all patients currently in the ED. At the end, the emergency physician decides whether the patient is ready to be discharged, or needs to be admitted to a hospital unit.


Delays happen when needs are far greater than the resources available to satisfy them. Even when the resources available are on average sufficient, delays will still happen when the needs are highly variable — that is, if they change from one person to next (as is often the case in healthcare), or from one interval of time (an hour or a day) to the next. One analogy is running on a flat surface as opposed to a constantly undulating surface. Even if the up-inclines equal the down-inclines, the energy you spend on the up-inclines exhausts you sufficiently to slow you down.

From one hour to next…

Consider the average number of patients who arrive in each hour of the day at a large hospital. 0 refers to the hour between midnight and 1 am, and 23 to the hour between 11 pm and midnight. Between midnight and 1 am, 7 patients arrive on average whereas during peak hours (from 10 am-8 pm) 14-16 patients arrive on average each hour. Remarkably, the shape of this curve is identical for all EDs across the country: the averages, in unison, simply shift upward or downward, depending on how large the ED is and where it is located.

Arrivals by Hour of Day

As always, averages don't tell the whole story. For example, the average between 4-5 pm may be 16 patients, but on 25% of the days the actual number of patients between 4-5 pm will be higher than 18; and on 5% of the days the actual number will be higher than 22. It's also possible that a higher than average number of these arriving patients are in serious condition, requiring more attention. So the ED staff and diagnostic facilities can find themselves overburdened; and it can be hard to catch up as queues from one hour follow you to the next. If the next hour also turns out to be a high volume, above-average hour, then waiting time can go up dramatically.

Whom to see next?

The magnitude of delays is also influenced by how doctors and nurses decide which patient to see next. In general, a first come first serve rule, which applies to most other services and which promotes fairness among those waiting, cannot be used in an ED. Just notice how absurd this would be: there is a patient critically wounded in an accident waiting, but just because she came a few minutes after someone with a sore throat, the emergency physician chooses to see the sore throat patient first!

So the sickest patient first is indeed the better rule. But the problem is figuring out who is really sick and who isn't. A person who has to wait because she has been deemed less sick may end up being sicker than most. This is because the triage nurse's initial assessment is not a perfect indicator and understandably prone to error. Even if the initial assessment is correct, there are limits to the number of patients the clinical staff can care for. In The Waiting Room, the man with the serious gunshot wound just has to wait because – as the triage nurse puts it – “we got so many in the same acuity”. A little later, as ambulances inform the hospital of arriving trauma patients, a doctor says:

“When we get the ring down from the paramedics that a trauma patient is on the way, literally 12 to 15 people stop whatever they are doing to focus on this one trauma. If several traumas roll in at the same time, that is a huge impact on the rest of the waiting room. People with legitimately serious illnesses get bumped over and over and over again for the trauma service.”

Elsewhere in the hospital…

Delays can be exacerbated by hold-ups in other parts of the hospital. In 2010, 13.3% of ED visits in the country resulted in hospital admissions. Admitted patients are moved from the ED to a bed/room in an appropriate inpatient unit so that they can be kept under observation. The average admitted patient stays 4-5 days at the hospital; some patients stay on for weeks or months, which skews the distribution considerably. If the inpatient units are at full occupancy, then the patient who needs to be admitted cannot be moved and continues to stay in the ED. In emergency medicine parlance these patients are called “ED boarders”. The presence of ED boarders — who still need to be monitored by clinical staff — in turn blocks access to care to those in the waiting room. If the problem gets sufficiently serious, the hospital starts diverting ambulances to other hospitals.


So what we have is an interacting network; a lack of flow in one part influences others. One way to reduce delays is to discharge inpatients in a timely fashion, so that their beds/rooms can be made available to other waiting patients. But this is easier said than done. I once attended a late morning meeting of nurses, doctors, and social workers as they went through their current list of inpatients with serious respiratory conditions. It was surprising how many patients had not been discharged simply because insurance had not agreed to pay for their next stage of care: transfer to a skilled nursing or rehabilitation facility. One patient was eligible for a Medicaid program in a neighboring state, but felt too sick and overwhelmed to fill the 21-page eligibility form. Such patients have no option but to stay on in the hospital, sometimes days or weeks longer until insurance issues are sorted out, occupying a bed that is also needed by other patients.


When all factors contributing to delays come together, it results in the kind of chaos that hospitals like Highland and a very large number of other urban EDs experience on a regular basis. All the tremendous technological advances in medicine that the US has made stand humbled when it comes to task of delivering care effectively.

What kind of solutions are there, if any? At the day to day level, there are questions on how shifts for physicians, nurses, technicians, transporters, and patient escorts should be scheduled to meet the demand which varies by the hour. Some of my research has focused on creating computer models of patient flow, using historical data obtained from hospitals. This approach mimics hospital operations as a random statistical experiment, and helps quantify the interplay between schedules and workflows in different parts of a hospital.

Such a mathematical take on the problem of delays, while useful and necessary, can only go so far. In large organizations, it is common to have groups compete for primacy and resources. Hospitals are no different. Surgeons have very different priorities compared to emergency physicians; nurses may have an entirely different view. Administrators may have their eye on financial and revenue issues for hospitals, and how policy changes at the state and national level might affect reimbursement. There is always concern about lower revenues if not enough inpatient beds are occupied; but at the same time, the basics of queuing theory make clear that when occupancies are over 90% significant delays are inevitable, and it is the patients who suffer.

With so many contradictory impulses, it isn't a surprise that hospitals do not function as an integrated whole.


Even if hospitals do get their act together, they can at best fix only a part of the problem. For a truly comprehensive answer, one has to look outside of hospitals, into the heart of the communities where the hospitals are located. What are the health problems that lead to so many sicknesses? Who is affected by these sicknesses? What are the socio-economic contributors? If hospitals and emergency departments provide very expensive, state of the art care, then why do patients keep returning? Can some of the conditions be managed in an outpatient primary care setting?

In 2011, Atul Gawande, writing for the New Yorker, drew attention to the work of Jeffrey Brenner, a primary care physician practicing in the town of Camden, New Jersey. In an interview with Gawande (video summary here), Brenner describes one example of how getting a lot of expensive hospital care does very little for the patient:

It was really obvious from the data that the most expensive people were getting terrible care. They were getting disorganized, fragmented and uncoordinated care.

And I knew them, so I knew what kind of care they were getting. And it was a very common scenario to have an extremely complex patient discharged from the hospital, show up in my office, and I'd walk in the exam room, and say: “Mrs. Rodriguez, I haven't seen you in three months. Where have you been?” “Well, I've been in the ICU for a month and a half. I've been in the hospital for another couple of weeks and just got out a couple of days ago.” And I said, “Well, what happened?” And she'd say: “Well, I'm not really sure. Lot of doctors came in the room. They never really explained anything to me. No one translated, and I'm not really sure. But I got this scar, and I've got this whole bag of medicine, and I've got a one-page carbon-copy discharge sheet, and all the meds have been changed, and [it] doesn't really say anything.”

So you'd call the hospital, and the hospital had not done the discharge summary yet. They hadn't dictated anything. They couldn't find the chart. You couldn't figure out who had actually taken care of the patient.

So now you've got this incredibly sick person who's got new complaints. They're complaining of chest pain today and shortness of breath. You really have no idea what happened to them in the hospital. They cost $20,000 while they were in the hospital, so the public paid $20,000 for their health care, and now I'm trying to figure out how to make sure they don't go back to the hospital, and I can't figure out what the hospital did to them.

Camden, a town of 79,000, has high crime rates and its public institutions have broken down. Brenner along with the town's help has set up an organization, called the Camden Coalition of Healthcare Providers, to proactively reach out to the vulnerable patients in the community who happen to be some of the highest users of emergency rooms, intensive care units and inpatient beds. The people who reach out to these patients are nurses and social workers. They might visit the patient's home to figure out what the problems are, and see if they could be addressed. In one case, they discovered that the serious respiratory problems the patient was suffering from possibly stemmed from the condition of his apartment – broken ceilings and inadequate ventilation. Here's another example:

Last November, a 55-year-old man with nine chronic conditions was admitted to a Camden hospital and assigned to a care manager at Camden Coalition…In the previous year, the patient visited the ED nine times and had six inpatient stays, racking up total charges of $312,000. He is a dual-eligible covered by Medicare and Medicaid, which paid a total of $59,000 for that care.

“We coordinated his home care, transport, meals, crutches, wheelchair, and dialysis. We got him to see a nephrologist, got him on the transplant list, and arranged numerous other services,” says Brenner. In the six months after the 55-year-old patient “graduated” from care coordination, he had had no ED visits and no hospitalizations. [Link]

Brenner's approach, as he himself acknowledges, is not new. At its heart is a stronger, considerably enhanced model of primary care, with an emphasis on prevention, getting to know the patient well, developing trust, bringing back the lost art of home visits, and advocating for the needs of the patient. There are similar initiatives underway all around the country, and have been going on for several years now. Camden Coalition just happens to be a striking and dramatic example. It took Brenner and his team many years of poring over hospital databases, investigating the role of crime, housing and geographic location in Camden to arrive at an understanding of health trends and needs in the town. This is inspiring grass roots work. The task of coordinating care for the sickest patients is quite challenging. Some patients welcome the personal attention provided by the nurses and case-managers reaching out to them. But not every patient responds the same way. Nevertheless preliminary findings suggest costs and hospital visits for such “high utilizers” have gone down by 40-50%.