What I've Learned As A Hospital Medical Director

Alright, from the title of this post, I’m sure you are trying to guess what it is. Open heart surgery? A craniotomy? Separation of conjoined twins? Wrong. The most dangerous procedure in medicine is the hospital discharge.

From the time a patient is admitted to the hospital, the medical resident or the hospitalist has one mission – to get that patient discharged, preferably alive. That’s because the discharge is the ultimate goal of a patient’s hospitalization.

We are singularly focused on discharge. As soon as a patient hits the nursing unit, we are pressing the admitting physician to enter a projected discharge date so that the hospital’s case management machinery can start working to get that patient out of the hospital. We analyze hospital length of stay for each physician, rewarding those who keep their length of stay index below one and chastising those with a length of stay index greater than 1. We put enormous resources into getting discharges out of their hospital room by noon so that we can make room for the next patient.

So why aren’t we patting each other on the back for every hospital discharge, after all, the fact that the patient is being discharged means that they are getting better so we were successful, right?

The problem is that there is so much that can go wrong when we discharge a patient. Here are just a few of them:

1. Medication reconciliation. Everybody talks about it but nobody does it very well. In order to keep our hospital costs down, we have a relatively limited number of medications in our hospital pharmacy formularies, limited to those drugs in each category that we are able to get the best pricing on. The problem is that these are not the drugs that the patient will be taking at home which are dictated by their insurance company’s formulary. And so the drug that the patient got better taking in the hospital is often different than the drug that the patient will pick up at the drug store. Even more of a problem, the patients frequently assume that they are supposed to take both the drug that is on their hospital discharge instructions and the drug that their primary care physician had prescribed for them prior to their hospitalization. I can’t even count the number of times that one of my patients who had been taking Advair for years comes into the office after a hospitalization taking both Advair and Symbicort because Advair wasn’t on the hospital formulary and the hospitalist put them on the equivalent Symbicort when they were in the hospital.
2. Medication access. This has gotten a little better as patients have had better access to health insurance under the Affordable Care Act. But it is still common for a drug to be prescribed at discharge from the hospital that the patient can’t afford once they go to pick it up at their local pharmacy. So what do they do when faced with a bill for a $300 antibiotic? They don’t buy it. And then they get sick again.
3. Nursing & doctor care. When they are in the hospital, patients are getting vital signs every 6 hours and seeing the hospitalist once or twice a day. As soon as they leave, that changes to seeing their primary care physician 3 weeks from now. We monitor patients up the wahzoo in the hospital and then release them into the wild the minute that they walk out the door.
4. Oxygen. This one is a pet peeve of mine, as a pulmonologist. We check the patient’s oxygen saturation by oximetry every day or multiple times a day in the hospital – almost always when they are resting quietly in bed. But they are not resting quietly at home, they are walking around, climbing stairs, carrying groceries, etc. And their oxygen saturation is a lot different when they are doing those activities. I’m always dismayed when a patient shows up in the office for their hospital follow up and the nurse checks their oxygen saturation right after they walk 50 feet from the waiting area to the exam room and the saturation is 76%. When you go back to the hospital record, it was always in the 90’s because it was always measured after the patient had been lying in bed for an hour.
5. Medical follow-up. If a patient gets discharged on a Saturday morning, the case managers (if they are even in on Saturdays) can’t schedule a follow-up office visit with the patient’s primary care physician. So instead, the discharge instructions will typically say something like: “Please call your primary care physician’s office on Monday to make an appointment to be seen within 3 days.” Here’s what usually happens: the patient forgets to call, the physician is out of town for 2 weeks, the physician doesn’t have any available appointments until December, or the patient doesn’t even have a primary care physician.
6. Communication with the outpatient physician. So lets be optimistic and the patient does call and get an appointment to see his primary care physician the next Tuesday. He shows up and his doctor asks him why he is there. The hospital policy is that a discharge summary has to be dictated within 5 days of discharge and the hospitalist hasn’t gotten around to dictating it yet. The patient says he was in the hospital because of high blood pressure. The primary care physician checks his blood pressure and it is normal so he sends the patient out. Two weeks later, when the discharge summary finally arrives in the mail, it says that the patient was admitted with hypertension and chest pain and was found to have an ejection fraction of 15%, an aortic dissection, and syphilis and that he should see his PCP for treatment.
7. Diet. No human being ever gets put on a regular diet in the hospital. It is usually some combination of low salt, carbohydrate limited, caffeine-free, diabetic no added sugar, 15 gram protein diet. Its no wonder they all hate our hospital food, its nothing like what they eat at home. And so the patient with heart failure goes home and what’s the first thing he does? Open up his pantry and pull out a jumbo bag of Doritos and then wash it down with a liter of original Coke. When you ask him about it the next day when he’s back in the emergency department, he’ll tell you quite honestly that the discharge instructions didn’t say anything about avoiding Doritos.
8. Pending tests. As soon as a patient is out of sight, they are out of mind. And so the chest CT scan result that was pending at the time of diagnosis comes back showing a “large lung mass that should be considered lung cancer until proven otherwise”. But the patient is gone so nobody sees the report. Or maybe the final culture from the bronchoscopy that the pulmonologist did comes back with Cryptococcus. But the report goes to the hospitalist who doesn’t know what Cryptococcus is and assumes it must be a part of normal flora.

There are a thousand things that can go wrong when a patient gets discharged. And that’s why the discharge is the most dangerous procedure in medicine.

October 31, 2016

Our hospital finished the last year with an inpatient mortality index of 0.54, a fantastic accomplishment. That means that we provide great care and we play the documentation game well. If the hospital mortality index is 1.0, that means that you had exactly the number of inpatient deaths that other hospitals have on average, after those patients are adjusted for their case mix index (CMI; a way of quantifying how sick the patients are). If your mortality index is > 1.0, then you had more deaths than the average hospital and if it is < 1.0, then you had fewer deaths than average. Another way of stating this is that the mortality index is the ratio of observed: expected mortality. At 0.54, our mortality index is one of the lowest of all hospitals in the country.

There are two ways you can keep your mortality index down: you can have few deaths (observed mortality) or you can document that you take care of a lot of sick patients. The best performing hospitals do both. As an example, your mortality index will be high if you have a young patient who came in for an elective cholecystectomy die (low expected mortality). On the other hand, if you have a 90-year old who has leukemia, is in heart failure, and is on dialysis come in for an emergent appendectomy and he dies (high expected mortality), your mortality index does not go up so much.

So if you want to be a best-performing hospital with a low mortality index, it is not good enough to just take great care of your patients, you have to document how sick they are. That’s where documenting secondary diagnoses that are present on admission becomes critical… and that’s where most physicians fall short. The hospital coders have to be able to pick those diagnoses out of the admission history and physical examination so it is necessary that the history and physical exam contain the precise words that indicate those secondary diagnoses that can significantly impact the case mix index. Words matter: you can’t just write “potassium = 3.0, will give KCl”, you have to actually write “hypokalemia, will give KCl”.

Here are the top secondary diagnoses that affect the expected mortality score:

1. Anemia (specific type of anemia and whether it is acute or chronic)
2. Acute respiratory failure
3. Coagulopathy (including use of anticoagulants)
4. Heart failure (systolic or diastolic; acute or chronic)
5. Chronic kidney disease (including the stage number)
6. End-stage renal disease
7. Diabetes (including whether it is type I or type II, controlled or uncontrolled, and what organs have manifested complications)
8. Hyperkalemia
9. Hypokalemia
10. Hypochloremia
11. Hypomagnesemia
12. Hypophosphatemia
13. Hyperphosphatemia
14. Hypocalcemia
15. Hypercalcemia
16. Hyponatremia
17. Hypernatremia
18. Dehydration
19. Hepatitis (A, B, or C; acute or chronic)
20. Liver disease
21. Protein calorie malnutrition (mild, moderate, or severe)
22. Metastatic cancer (including what organ it metastasized to)
23. Decubitus ulcer (including the stage number)
24. Pleural effusion
25. Pulmonary edema
26. Neurological or brain/spinal conditions
27. Sepsis
28. Shock
29. Transfer from an acute care setting
30. Requiring mechanical ventilation

We were all trained in medical school that the history and physical examination was all about our diagnostic impression and medical management. But the H&P is additionally an integral part of the documentation game. In the documentation game, the physicians are the players for the hospital and the physician gets more points for the more secondary diagnoses he or she documents. The winner is the hospital that documents that its patients are really, really sick and then discharges them alive.

October 29, 2016

Yesterday, a man was shot a couple of miles from our hospital. Not an unusual event, people get shot every day in America’s large cities. Our hospital is not a designated trauma center but we still have gunshot victims dropped off at our front lobby by their friends – either they die in our ER or we stabilize them and send them to one of the regional trauma centers in Columbus. Guns are a way of life in our city, as they are in all cities in the U.S. We require all patients to go through metal detectors to get into the emergency department – as do all hospitals in town – and finding guns is not uncommon at the metal detector. We require anyone with a weapon to hand it over to our security personnel until they leave – a lot like an old west saloon.

Shootings and murder are commonplace in the United States. We kill each other more than any other first world nation and we are right up there with third world countries where a murder doesn’t even make headlines. We’re pretty good at the detective work that goes into solving those murders, our problem is that we just have too many of them.

 

To kill a person, two things have to happen. There has to be someone who wants to kill someone else and they have to have an effective way of killing. So, if killing another person requires both intent and an available method, which of these is the reason for the United State’s exceptionally high murder rate compared to peer countries? I’d like to think that it is not because Americans are full of more hatred than other industrialized nations.

A gun is a highly effective way to kill someone and we have too many of them. Guns are everywhere in the United States. We have more guns per capita than any other country on earth. In fact, we have more guns than we have people in our country.

What is the answer? I’m not any smarter than anyone else. But solutions need to start with what we agree on instead of starting with what we disagree on.

I think everyone in the U.S. would agree that a person shouldn’t be able to walk into a sporting goods store and by a nuclear weapon. But almost everyone would agree that a person who wants to go duck hunting should be able to buy a shotgun.

Everyone would agree that someone who was previously convicted of felony armed robbery and is also a card-carrying member of a terrorist group should not be allowed to buy guns. But almost everyone would agree that a law abiding citizen with no criminal record should be allowed to buy a gun.

We agree with the extremes of gun control. What we disagree with is where the line needs to be drawn in the middle. By spending all of our political energy on fighting each other in courts and legislatures about exactly where that middle line is for gun control, we’ve accomplished almost nothing. If we instead started with the fringes of the gun control debate and tried to figure out on what we can agree on, we might actually get something accomplished.

Don’t get me wrong. I’m not a raging liberal out to pry guns from everyone’s hands. I used to be a member of the NRA (when I perceived it as being an organization primarily dedicated to gun safety and recreational hunting rather than a lobby group for gun ownership). Some of my fondest memories from childhood were bird hunting with my father in early winter in rural Southern Ohio. And I’ve personally annihilated hundreds of clay pigeons over the decades (with permanent tinnitus as evidence).

But having too many guns and putting them in the hands of the wrong people just creates a lot more work for all of us who keep America’s hospitals working. As it is right now, we just make it too easy to kill each other in our country. For most of Americans, the topic of gun control is polarizing. But if you work in an emergency department or a trauma operating room, the lack of gun control is just exhausting.

Sorry to be up on a soap box about this. But we have to make steps to stop our country from being the murder capital of the Western World. I would much prefer my hospital to focus on treating disease than treating the consequences of motive + opportunity.

October 18, 2016

On average, about 1 out of 5 patients admitted to the hospital with COPD in the U.S. get readmitted within 30 days. There is a wealth of medical literature analyzing COPD readmissions. For example, we know that patients at risk for readmission include:

1. Patients without physician follow-up within 30 days of discharge
2. African Americans
3. Older patients
4. Divorced/widowed patients
5. Those with longer initial hospital stays
6. Patients in nursing homes
7. Patients with anemia (hemoglobin < 8)
8. Patients with renal failure
9. Patients receiving cancer chemotherapy
10. Patients with low health literacy
11. Patients on Medicaid
12. Patients taking > 5 different medications

There are a lot of reasons why they get readmitted. Some of the reasons include:

1. Insufficient outpatient follow up
2. Medication errors
3. Poor transfer of information to primary care providers
4. Inability to pay for medications
5. Inadequate transportation
6. Incorrect inhaler technique
7. Lack of a “rapid action plan”

Interestingly, only 28% of patients with an initial hospitalization for COPD are readmitted with COPD. More than half of those readmitted are for non-respiratory conditions such as heart failure, arrhythmias, intestinal infection, sepsis, and electrolyte disturbances. Also, readmissions occur pretty quickly with 50% of readmissions occurring in the first 2 weeks after discharge. So what can we do to reduce COPD readmissions? Successful strategies employ interventions both during the hospitalization and after hospitalization.

During hospitalization:

• Screen patients for readmission risk factors
• Communicate with primary care providers
• “Teach back” to educate patient (respiratory therapists can be valuable)
• Use interdisciplinary clinical teams
• Start on maintenance long-acting inhalers
• Start on roflumilast (if indicated)
• Discuss end-of-life wishes
• Comprehensive discharge planning
• Ensure patients can get and manage their medications
• Schedule an outpatient follow up visit
• Make sure patients have a nebulizer

After discharge:

• Promote self-management (emergency action plans)
• Follow-up phone calls
• Develop and use transition clinics
• Home visits for patients with transportation barriers
• Electronic medical record management of information
• Establish community networks
• Use telemedicine when appropriate
• Arrange spirometry testing
• Enroll in pulmonary rehab
• Smoking cessation

We started a nurse practitioner-run pulmonary transition clinic at our hospital and were able to reduce COPD readmissions to 12.5%. However, we did see a sizable no-show rate and those patients who failed to show had a very high (27%) readmission rate. The specific actions that occur in our transition clinic include:

• Clinic appointment within 5 days of discharge
• Assess response to treatment
• Follow up lab and radiology tests
• Arrange pulmonary function tests
• Medication reconciliation
• Refer to indigent patient medication assistance programs
• Arrange pulmonary rehabilitation
• Smoking cessation
• Insure correct use of inhalers

Readmissions cost all of us in the long run since they increase insurance/Medicare/Medicaid costs. Care coordination and education are key elements of any readmission reduction strategy. Respiratory therapists are often in the best position to champion patient education in the hospital. Ultimately, it requires a culture change in our approach to COPD – culture always trumps hospital policy.

October 14, 2016

It is flu shot season and my goal each year is to give more influenza vaccines in my clinic than any of the other pulmonologists. So, I offer it to all of my patients and continue to be amazed at how many of them decline because “Every time I get a flu shot, I end up getting the flu”. There is no live virus in a flu shot so you are just as likely to get the flu from a flu shot as you are likely to get pregnant by taking a birth control pill.

So why are patients so sure that they’ll get an infection from the flu shot. There are two main reasons. First, they may have had some muscle pain at the injection site or even some mild myalgia after a previous injection – this is a reaction to the vaccine and not an infection. If anything, it means that the vaccine is working because your immune system is mounting a response to it.

The other reason patients think that they get the flu from a flu shot is from superstition. The average American gets 2-4 upper respiratory infections (“colds”) per year. Lets just say it averages out to 3 colds per year. That works out to 1 cold every 17 weeks. In other words, statistically, 1 out of 17 patients will get a cold within a week of getting a flu shot purely by chance. Because they associate that cold with the flu shot, they incorrectly deduce that the vaccine caused the cold (which they equate to the flu). By the same argument, 1 out of 17 patients will get a cold within a week of Easter but you don’t hear patients telling you that they got the flu from the Easter bunny.

As it happens, if it wasn’t for influenza, I never would have been born. My grandmother’s first husband was one of the 21 million people who died of the “Spanish” influenza epidemic of 1918-1919. She then remarried to my grandfather so if her first husband hadn’t died of the flu, I wouldn’t be writing this post now. In the United States, about 23,000 people die of influenza each year; some years more and some years fewer, depending on the specific strains that go around that year.

It is particularly important for all healthcare workers to get vaccinated so that they don’t become a vector to transmit influenza to vulnerable patients. A few years ago, I admitted one of my patients with pulmonary fibrosis to the hospital with worsened shortness of breath. On admission, I did a bronchoscopy and sent PCR testing for influenza – it was negative. We determined that he was in heart failure and he improved over the days with diuresis. He lived alone and had no relatives so during his hospital stay, he had no visitors. After about a week, he became suddenly worse with hypoxemia and high fever. I repeated the bronchoscopy and this time, his influenza PCR was positive for influenza A. Based on the incubation period, he had to have acquired his influenza in the hospital. Since he didn’t have any visitors, he had to have acquired it from one of the doctors, nurses, or therapists. He never made it out of the hospital and died of his influenza in our ICU.

So, I’m pretty passionate about getting everyone who works in the hospital vaccinated for influenza each year. I don’t care so much whether they get influenza but I don’t want them transmitting it to a patient who would be more likely to die from it.

October 12, 2016

Two days ago I got an urgent email from one of our case managers asking if I could see a patient with asthma in the office ASAP. She was in the hospital with her third asthma exacerbation in 2 months and the case manager was trying to keep her from being admitted yet again. Although my practice is primarily interstitial lung diseases, I do try to help out with other pulmonary diseases when I can. I pulled her up on our electronic medical record and found that she had never had an inpatient pulmonary consult with any of her previous asthma admissions.

So, why in the world would a patient with a disease that is this difficult NOT have a consult by a specialist? I think it is a reflection of how residents are often trained. Going back to my own internal medicine residency, the mantra of the senior residents was that “A consult is a sign of weakness”, meaning that if you ordered a consult to an internal medicine specialist, it meant that you weren’t smart enough to handle that problem yourself. It wasn’t just my own specialty of internal medicine – a senior surgery resident told me that he didn’t get medical specialty consults because “A surgeon can do anything an internist can do plus we can operate”.

Over the years, those attitudes have become embedded in the culture of American medicine. What we forget is that the practice of medicine is a skill and like any skill, if you want to be good, the practice of medicine takes practice. Let me give you the analogy I use with my medical students. If you want to learn to ski, you can’t just read a bunch of books about skiing. You can’t just watch a bunch of videos of experienced skiers. And you can’t just strap on skis, go to the top of the slope and then try to get to the bottom over and over again. Truly learning a skill takes all three: didactic education, observation of skilled practitioners, and practice on your own.

In some residencies, the didactic education will be awesome, for example, many academic medical centers. In some residencies, there will be extraordinarily experienced clinicians that you will observe, for example, many community hospitals. In some residencies, the interns and residents are often on their own and get a lot of “sink or swim” experience taking care of patients independently, for example, some VA hospitals. Each resident needs a different optimal mix of didactics, observation, and practice to reach his or her potential. But all residents need each of them in some combination.

The thing that sets a specialist apart is that he or she has had a lot more practice with a specific group of diseases. And that practice can translate into more nuanced care, particularly for those difficult-to-treat cases. Furthermore, the specialist that sees the patient in the hospital is often the one who will be seeing that patient in the outpatient clinic and that translates into better continuity of care.

The literature confirms that inpatient consultation improves outcomes in difficult cases. A study presented at the Society of Hospital Medicine showed that inpatient cardiology consultation reduced the 30-day readmission rate for heart failure from 26% to 15%. A study of inpatient geriatrics consults showed a reduction in 6 & 8 month mortality rates. A study of inpatient palliative medicine consults showed a reduction in readmissions from 15% to 10%.

As a specialist who sees both inpatient and outpatient consultation, I don’t believe that there is any question too small or medical problem too minor and so I will never criticize another physician for requesting an inpatient consult. However, I will criticize a physician for not getting an inpatient consult when they should have. A patient with a medical illness who is being repeatedly readmitted is more complex and takes a lot more of the consultant’s time than a patient with the same disease but more minor symptoms who never gets admitted to the hospital. Insurance companies recognize that, for example, a level III inpatient new/consult visit has 3.86 work RVUs whereas a level V outpatient new/consult visit has 3.17 work RVUs. The implication is that it takes more time and more complexity to take care of that problem if it is bad enough to result in a hospitalization than if the patient is in an ambulatory office setting. So, I don’t want to see that really sick patient in the office, I want to see them in the hospital where I will have more time to do the evaluation and more diagnostic and therapeutic resources immediately available.

An inpatient who is challenging enough that they need to be seen by a specialist immediately after discharge is challenging enough that they should be seen by that specialist while they are still in the hospital. Because a consult is not a sign of weakness.

October 8, 2016

Ohio State is about to expand its 340B program to include a free-standing pharmacy and the outpatient infusion centers. It gave me a chance to brush up on what a 340B program is. The 340B program was created by the federal government in 1992 as a way to provide discounted outpatient drug pricing to healthcare institutions that care for the poor. There are 6 categories of hospitals that are eligible that largely have in common that they are tax-payer funded to care for low-income and uninsured patients:

1. Disproportionate share hospitals
2. Children’s hospitals
3. Cancer hospitals exempt from the prospective payment system
4. Sole community hospitals
5. Rural referral centers
6. Critical access hospitals

In addition, hospitals have to either be state/government-owned, be a private not-for-profit hospital that has been granted governmental powers, or be contracted with the government to provide care to low income patients. In addition to hospitals, there are certain outpatient clinics that are also eligible to participate. As of 2014, there were 2,140 hospitals in the program, 90% of which are either critical access hospitals or disproportionate share hospitals.

The way the program works is that the Health Resources and Services Administration (HRSA) sets the maximum amount that drug companies can charge for outpatient medications – on average this is about a 22.5% discount. Medicare part B covers some outpatient medications (eg, cancer chemotherapy and rheumatoid arthritis infusion drugs); however, the hospitals participating in the 340B programs get paid the same from Medicare part B for these drugs as they would if they were not in a 340B program. Therefore, the hospital stands to make money on 430B drugs. On the other hand, drug manufacturers have to sell the hospitals their drugs at the discounted 340B price and so they would like to limit 340B programs so that they can have a higher profit.  All told, 340B programs save about $4 billion per year in drug costs. There are about 7,000 different drugs in the 340B program.

Ideally, hospitals participating the 340B programs use the increased margin that they get from the 340B programs to help support the care of lower income patients. For example, using profits to pay for rheumatoid arthritis infusion drugs for patients who are low income and have no health insurance and otherwise would not be able to buy these rather expensive medications. The danger is that there is the potential for some hospitals to expand chemotherapy and infusion clinics since they can make a higher margin on the chemotherapy and infusion drugs. Overall, 340B sales account for about 2% of total drug sales in the United States so it is not an enormous amount but 340B pricing is disproportionately affecting high cost chemotherapy and rheumatology biologic medications.

I can see both sides of the argument for the 340B programs but at least for our hospital, it will allow us to treat patients who previously were too poor to be treated for conditions like rheumatoid arthritis and inflammatory bowel disease in the past.

October 5, 2016