Editor's note: Dr. Aiken is Claire M. Fagin Leadership Professor of Nursing, Professor of Sociology, and Director of the Center for Health Outcomes and Policy Research at University of Pennsylvania. She is generally considered the nation's foremost expert on health policy as it relates to the nursing workforce. We spoke with her about how nurse staffing and the work environment can affect patient safety and outcomes.

Dr. Robert M. Wachter: What got you interested in the size of the nursing workforce and the way nurses were trained as potential predictors of outcomes? How did you approach those issues, and what were your key findings?

Dr. Linda Aiken: Early health services research that showed significant differences between hospitals in surgical mortality and pointed to hospital organizational characteristics that seemed to explain some of that difference in patient outcomes got me interested. These papers focused on things you could not change such as ownership (whether for profit or nonprofit), size, teaching status, and technology availability. Even though the earliest studies found a relationship between differences in nurse staffing and mortality, it was never pursued; yet, it was obvious that nurse staffing had something to do with outcomes. We decided to study hospital patient outcomes and make it our primary interest to analyze how much of the variation in patient outcomes was explained by differences in nurse staffing—was it causal or not. Our research shows that variation in nurse staffing and the educational qualifications of nurses are major factors in explaining variations in patient mortality and essentially every other adverse outcome that patients experience.

RW: Has any study looked at whether intentional variations of nurse staffing—comparing a place that had poor staffing to one with better staffing in a trial design—made a difference? Or has it all been observation of what happens in different settings that happen to have different nurse staffing?

LA: There haven't been any trials, and you can imagine how difficult it would be to convince hospitals to make a change in nurse staffing that might cost millions of dollars. So, we've been creative and tried to use natural experiments. One of the best natural experiments occurred when California enacted mandated nurse-to-patient ratios. When it was implemented on January 1, 2004, the hospitals that were not in compliance to the staffing ratios had to change on that day and they did. Our research has shown that staffing did change substantially in California hospitals—even in safety net hospitals, which have been very difficult to get to change on hospital nurse staffing.

Almost 15 years later, California still has the best nursing-staffed hospitals in the country. The state has seen steeper declines in mortality and improvements in other indicators than other states. It has been a little easier to look at the causal relationships between improvement in nurse education, patient outcomes, and quality of nurse work environments, because those do change over time. We have shown that hospitals that improve work environments and increase the proportion of nurses with baccalaureate degrees have substantially faster reductions in mortality, surgical and common medical reasons for hospitalization, and other outcomes like readmissions, poor glycemic control, and negative resuscitation outcomes.

RW: What are the elements of the work environment that turned out to be important?

LA: We have a global measure of the work environment to measure staffing adequacy, which is different from what the actual staffing is—it's the appraisals of tens of thousands of nurses that work in hospitals. So they are appraising the adequacy of nurse staffing in the environment in which they work. Doctor–nurse relationships is another factor. Investment in an educated clinical staff is another, and an institutional commitment to quality and patient safety is another.

Our work was included in the IOM report and one of the major reports in the series was Keeping Patients Safe: Transforming the Nurse Work Environment. The basic thesis of that report was that it would not be possible to keep patients safe unless the quality of the nurse work environment was substantially improved.

Since the IOM report, we have been following up to see whether any evidence shows that the culture of patient safety actually has improved. We have empirical evidence through 2016 that suggests the culture of patient safety in hospitals has not changed enough. We hypothesize that this explains why we haven't achieved the gains in reducing patient harm that the IOM was aiming for. Although researchers debate about how much improvement we've made in patient safety, I think everybody agrees that it is not what we anticipated that we wanted to make.

RW: Why do you think the balance has not shifted sufficiently so that people read your work and just go ahead and increase their investment in nursing?

LA: Some hospitals are improving their work environments. We estimate that slightly more than 1-in-5 hospitals have substantially improved their clinical work environment since the IOM report. But it is disappointing that in the majority of hospitals there has been no change, and some hospitals have showed substantial deterioration in their work environments over that time.

So why haven't they improved work environments? It seems more difficult to make some of these organizational changes than it does to look for the magic bullet that might prevent patient harm. Our take on the safety interventions that have come forward so far is that they have been low-hanging fruit, versus organizational reform that the IOM called for. Checklists, care bundles, hand hygiene, and rapid response teams all seemed doable. But in the context of those interventions being tested and many of them showing promise, over the same period there has been decreased professional autonomy and increased burnout among both doctors and nurses. We've added a lot of management levels and we have a whole bureaucracy of safety people, but our data suggests that little has changed about the fundamental organization of clinical care, particularly in hospitals.

We still have chaotic environments and frequent organizational failures. We think one reason why safety interventions that have been shown to work under controlled circumstances don't work in practice is because we haven't fundamentally changed the safety of the clinical environment, and we're superimposing these interventions on top of a defective environment. One example would be care bundles for central line infection prevention. Research suggests that these care bundles work, but if you look at the small print it says if they're implemented 95% of the time. It would be virtually impossible to adhere to any guideline 95% of the time, because many hospitals don't have enough staff, and chaotic work environments prevail.

RW: Focusing on the sufficiency of nurses and their training, you could make an argument that what California did was right, so it should be mandated that you have enough nurses. Why has uptake been relatively slow?

LA: Well, it's clear that we have a huge variation of staffing across hospitals. Almost everything about nursing varies across hospitals. And these differences line up very closely to the variation across hospitals in mortality, infection rates, and other adverse patient outcomes. But other than the mandated ratios, there hasn't been too much of an incentive for hospitals to improve staffing. I'm not even sure that management fully recognizes how important nurse staffing is. They tend to think of it as a flat-out cost to be minimized, when research shows that it's costing hospitals a lot not to have enough nurses.

For almost every national issue that hospitals are interested in or that are part of Medicare's value-based purchasing, staffing is associated with them. Readmissions, infection rates, never events like poor glycemic control—they're all directly related to nurse staffing. Those are all very expensive too. Not having enough nurses is not saving anybody any money.

But you asked why hasn't there been more of an uptake. You live in hospitals as much as I do, and people are there from different disciplines with different expertise and different outlooks. My own experience is that nonclinicians, who are making a lot of decisions in hospitals, don't naturally gravitate toward understanding clinical outcomes. They're more interested in financial outcomes and don't appreciate that the two are always linked together. So that has been a hard sell. Researchers tend to be interested in clinical outcomes: We want to reduce harm to patients and mortality. We tend to focus on clinical outcomes without making the business case for why these improvements are also going to affect the bottom line of hospitals. The business case for nursing has improved in recent years.

RW: Tell us about current research on what nurses think about patient safety.

LA: We look at nurses from a sociological perspective, which would argue that the best way to understand complex organizations is to study the people that are part of the organizations and not necessarily ask the executives—or in addition to asking the executives. So we study nurses as informants of everything going on in hospitals and other clinical settings because they are at the nexus of patients, doctors, and management.

In our study undertaken in 2005 to 2008 of more than 22,000 nurses in 577 hospitals, we found that 38% of nurses gave their hospitals an unfavorable grade on patient safety. In our newest study undertaken in 2015 to 2016 (currently under review), about one third of nurses still gave their hospital an unfavorable patient safety grade. A large proportion of nurses would not always recommend their hospital to family or friends that needed health care. Then we used AHRQ measures on patient safety culture to appraise our hospitals. A majority of nurses say that they and other staff feel like their mistakes are held against them and one third report they're frequently interrupted due to missing supplies, broken equipment, missing medication. That is not the culture of patient safety that the IOM recommended.

Then we looked at the huge variation in nurse staffing across hospitals and in the proportion of nurses with BSN [bachelor of science in nursing] degrees. Even though the IOM recommended that 80% of all nurses be qualified at the BSN level by 2020, in our recent data, plenty of hospitals still have only 10% or 20% of their nurses with BSNs. We also saw huge variation in nurses' ratings of their practice environment, with around 40% of all nurses reporting that their environments were only fair or poor. Again, we have validation that there is tremendous variation in the quality of the work environment and plenty of studies that show that variations in these nursing factors are associated with mortality and other adverse patient outcomes.

RW: Let me turn to a couple of emerging issues. When you started doing this work, the nurses all worked for the hospital but the doctors didn't. I imagine that the politics and the economics of that influenced your work, in that your audience in some ways was hospital leadership. Now, more and more physicians work for the organization. So the dynamics are becoming more similar in terms of the relationship between management and doctors and management and nurses. Does that change the nature of your work, or are there lessons that physicians should be gleaning from your experience?

LA: There is new interest in burnout among physicians. We've been studying burnout in nurses for 30 years. The same organizational factors that lead to burnout in nurses are bound to lead to burnout among physicians, such as limited autonomy and control over your own work, which is happening to doctors since many more of them are now employees. Doctors are now beginning to see through nurses' eyes how difficult it is to change some of the operational issues such as poor staffing, health professional fatigue, and operational failures that they didn't fully appreciate before, because they were not as burdened with them as they are now. In some ways, it's good for doctors because now they have more of an appreciation of organizational deficits and can hopefully join nurses in trying to create safer work environments for patients and clinicians.

RW: The majority of nurses in hospitals are working with and through technology. How has it made life better or worse? Are new challenges emerging as everyone's work life becomes much more intertwined with computers?

LA: I would say very definitively that technology has increased the work burden and need for more nurses, and that is one of the reasons why hospitals perceive that they cannot catch up in terms of hiring enough nurses. All the technologies introduced are nurse-intensive. I cannot really think of a single example of any technology, especially in hospitals, that has saved any nurses any time, including electronic health records, which are very physician time-intensive too. Hospitals with better work environments get better outcomes from their electronic health records than hospitals that have poor work environments. But the main issue is that new clinical and information technologies are all nurse-intensive. Nurses like technology and it improves care, but more technology along with the ever-shorter length of stay is a relentless burden on nurses in terms of increased workload without further increasing the nurse staffing.

RW: One would have thought that in every other industry, you bring in technology and you need fewer people and less time, but that is not the way health care works.

LA: No, and nonclinical managers don't really appreciate that either. It's all very labor-intensive.

RW: So 10 years from now how is all of this going to look?

LA: Hopefully researchers will continue to provide evidence to justify a business case for investments in nursing. An example is our study showing that mortality for matched surgical patients is significantly lower in hospitals with good nurse work environments and care costs the same or less than in hospitals with poor environments. In addition to trying to create this business case to finally get the attention of nonclinical management, we're working on the idea of causation. With a lot of competition for resources, I can see why managers would want to know if they make an investment in x that y is going to happen. If you buy a new piece of equipment, you could probably estimate what your revenues are on the new piece of equipment, but it's more difficult to do that with human resources. We are increasingly focusing on examining panels of hospitals over time to determine the extent to which those that improve nurse work environments and move toward a workforce with a higher proportion of bachelor's educated nurses experience better outcomes. We think evidence suggests the way to reduce patient harm as recommended by the IOM is to dramatically change the way hospitals are organized from a clinical perspective.