From nurse education, clinical, or workforce issues through to healthy living, Nursing Review has it covered.

A nursing mantra for much of the past two decades has been evidence-based practice. US nursing researcher Dr Karen Monsen believes it’s time to rethink that mantra and instead start mining ‘big data’ for practice-based evidence of expert nursing. Fiona Cassie reports.

Every day, millions upon millions of pieces of health data are electronically recorded and stored as part of health systems around the world – much of it by nurses.

“It’s inconceivable really now how much data we have,” says Karen Monsen, a US-based nursing informatics expert. She believes nursing should be tapping into and analysing this ‘big data’ for evidence about nurse-sensitive and nurse-delivered care and her research agenda is to find the best methods to do just that.

Associate Professor Monsen is a self-described ‘mild-mannered public health nurse’ who discovered nursing informatics after being thrown into it at the deep end in the late 1990s and told to computerise her service’s nursing records. One PhD later, she is now co-director of the University of Minnesota’s Center for Nursing Informatics and is also a director of the Omaha System Partnership (see box).

Speaking in her keynote address on ‘big data’ at October’s National Nursing Informatics Conference, Monsen questioned whether nursing should be so won over by the concept of evidence-based practice (EBP).

“We have swallowed it [evidence-based practice] hook, line and sinker,” she said.

The past two decades have brought the Cochrane Collaboration, Joanna Briggs Institute and, until recently, the New Zealand Guidelines Group all promoting and offering evidence-based health care. “We believe in this; we strongly want to do the right thing for our patients,” said Monsen.

But Monsen also said that recently there has been more and more discourse over whether EBP is enough and whether it is truly serving nursing well. She questioned whether expecting expert clinicians to always consult EBP guidelines may be seen as devaluing or diminishing sound clinical judgement. She also raised issues around the limitation of the evidence from which EBP-based guidelines are drawn up.

“Are they [guidelines] developed with the entire population in mind or just certain individuals or participants in that research?” she said. “And what about other biases of evidence – like the controls, or the environment? How many nurses actually work in ‘controlled environments’?”

 

Nursing nuances need recognition

Monsen said nurses need evidence that reflects the real world “with all its complexities and nuances” to be able to provide the best personalised care for individuals.

“There is this assumption that there is no data source from which we can begin to understand the clinical expertise that comes with that expert clinician who is working with that unique individual in the community. That assumption is false.”

An alternative approach was to delve deep into ‘big data’ to find practice-based evidence as a complement to evidence-based practice and as an “essential component of nursing knowledge”. (See definitions and ‘Big data’ sidebar.)

She said the bottom line is “let the data speak”. One way big data research can do this is by reversing the conventional research concept of testing hypotheses and instead use the data to generate hypotheses. 

“Once that data starts speaking to us – about what we should examine or what we should know – then we can start testing those hypotheses using traditional statistical methods,” said Monsen. “And what we get back is the voice of nursing and the voices of patients that have never been heard before… practice-based evidence.”

Monsen shared a number of examples of practice-based evidence developed by mining patient care data recorded by public health nurses on computers (using the Omaha system of reporting and documenting their nursing work with clients in the home; see more in ‘Omaha’ sidebar).

One ‘bucket of data’ involved high-risk young mothers and babies who were visited in their homes by public health nurses. The aim was to analyse the data to find out what explained the variability in outcomes for the young mothers the nurses worked with. Monsen said the data showed that the patient herself (50 per cent) and her existing problems (17 per cent) together explained two-thirds of the variability in changes the young mothers were able to make over time. The remaining third was influenced by the individual nurse who worked with the mother (17 per cent) and the interventions (17 per cent) that the nurse initiated.

“So it is critical for us nurses to always be at our best – and it is critical that we do the right thing,” she said. “The implication for research here is that we need to incorporate who the nurse is into all of our models as an important part of the research.

“And for policy we need to make sure we are taking care of our nursing workforce.  We need the best nurses; we need the best fit between the nurse and the assigned patients; we have to make sure we are taking care of our workforce.”

Another study explored further the concept of individual nursing contribution by analysing the data to find the patterns in individual public health nurses’ care of  individual patients. This analysis was depicted graphically by using colours for individual patient problems; colour shadings to show the nursing actions in response; height to show how frequently nurses delivered these actions; and horizontal length for how long the nurse was involved with the client.

The result was 403 rainbow images (looking a little similar to a geological cross-section of the sea bed or a mountain range), which experts classified into 29 distinct patterns. “No chart is the same,” Monsen pointed out. The colourful graphs demonstrated clearly that nurses’ care plans were unique for each individual patient and also started to reveal ‘fingerprints’ of individual nursing styles.

Monsen said the advantages of working with large datasets include being able to examine evidence around relatively rare situations – like a study she did focusing on mothers with intellectual disabilities, in which she found 17 matches. Analysing the data for these 17 mothers revealed they had twice as many problems as mothers without intellectual disabilities, received twice as many public health nursing visits and interventions, and responded by showing improvements in all problem areas – though did not reach the desired health literacy benchmark for parenting.

Other big data analyses she cited revealed hard evidence that frail older people were more likely to be hospitalised if they didn’t get enough skilled nursing intervention, plus evidence that public health nursing visits had a positive impact on the health literacy of immigrants and refugees and reduced health disparities between ethnic groups.

“This is why you [nurses] need data so you can say ‘I make a difference’; so you don’t have to wait for somebody to discover you in a research study and say you make a difference. We all have this power.

“For all these studies we can look at our interventions and our outcomes. And we have been able to explain this notion of practice-based evidence; and that we should encourage personalised interventions by expert clinicians to tailor and meet individual needs.”

It all begins with the daily nursing task of entering notes on a computer in a systematic way that allows the nurses’ friend, ‘big data’, to in time reveal what impact nursing interventions make, or don’t make, on patient outcomes. Or, to reiterate Monsen’s advice: “Nurses: let the data speak!”

 

Definitions*

• Nursing practice: What expert nurses know and do every day to ensure wellbeing and safety of patients: in the real world; and for unique patients and situations.
• Practice-based evidence: How does adding X intervention alter the complex personalised system of patient Y before me?
• Big data: Large datasets of structured or unstructured information that may require new approaches for analysis. 
• Practice-based data: Data from nursing assessment and documentation that is part of routine nursing care (and inputted into computers).

*Definitions as used in Karen Monsen’s presentation and drawn by her from a variety of sources.

Big data research in nursing

• Using large data sets to examine important healthcare quality questions.
• Looking for hidden patterns in the data.
• Hypotheses generating vs hypothesis testing.
• Can create new voice for nursing and patients by revealing practice-based evidence.

N.B. Recommended further reading about big data research: the free access online book The Fourth Paradigm: Data-Intensive Scientific Discovery published by Microsoft Research.