AANS Neurosurgeon | Volume 28, Number 3, 2019


The Science of Practice and Vascular Neurosurgery

Print Friendly, PDF & Email

In 1938, Harvey Cushing and Louise Eisenhardt published Meningiomas, Their Classification, Regional Behaviour, Life History, and Surgical End Results.1 This iconic text demonstrates the Science of Practice algorithm elegantly. The product of more than 20 years, it demonstrates the habitual and systematic collection, analysis and interpretation of data and then application of the analysis to improve care. These are the basic tenets of the Science of Practice algorithm. Today we may collect the data in digital databases, analyze it with more sophisticated biostatistics and publish it online, but the quintessential components of the Science of Practice remain the same.

On a very personal level, such a process has changed my practice of carotid surgery. I also suggest a way to apply this powerful clinical research technique in the study of unruptured aneurysms.

Carotid Endarterectomy: My Journey

I was taught to perform carotid endarterectomy (CEA) using general anesthesia with electroencephalographic monitoring and postoperative care in the intensive care unit (ICU). Patients routinely had intra-arterial lines placed and stayed in the hospital for four or five days. However, a review of my early experience2 defined the kind of medical co-morbidities existing in these patients and demonstrated the scope and timing of complications that occurred. More importantly, starting in 1990, I began to apply the Science of Practice algorithm by entering patient characteristics, processes of care and outcomes into a prospective observational database (POD). Over the course of my career I collected information on more than 2,000 consecutive CEA patients. The analysis of this data has allowed me to generate new knowledge about my practice, to better inform my patients and to improve the value of my surgical care for patients with carotid artery disease.3

For example,

In the late 1990s I instituted dual antiplatelet therapy with aspirin and clopidogrel started before and continued for six weeks after CEA. The use of regional anesthesia had reduced non-neurological complications but had not reduced the perioperative stroke or death rate. The institution of dual antiplatelet therapy demonstrated a greater than 50 percent reduction in perioperative stroke and further reduced perioperative myocardial infarctions. A recent review found that the perioperative stroke and myocardial infarction rate are now less than one percent and that the cost to deliver surgical care for my patients is 33 percent less than for other patients undergoing CEA at our institution.7

Beyond CEA: What’s Next?

The Science of Practice algorithm is perfectly applied to multi-center studies to generate reliable clinical research data when randomized controlled trials (RCTs) are not feasible or will result in non-generalizable data.8 Most neurosurgeons recognize that in the ideal situation RCTs have advantages over other study designs. Fewer recognize that the lack of masking, lack of equipoise in treating surgeons, unrepresentative surgeons and patients, the effect of evolving technology and other factors often adversely affect the reliability and generalizability of surgical RCTs.8  

The Science of Practice may be a better option than RCTs particularly for many neurosurgical problems such as comparing endovascular or open surgical treatment to conservative management for patients with unruptured aneurysms.9 Applying the Science of Practice algorithm requires the design of a POD that generates reliable causal inferences to generate the best treatment option. This necessitates the application of sophisticated statistical analyses such as propensity score matching to the data in the POD.10 To accomplish this we need to answer the following questions when designing the POD:

  1. What is the hypothesis we are testing?
  • Invasive treatment of unruptured aneurysms improves the functional health status of patients compared to observation alone.
  1. Who are the decision makers for treatment assignment?
  • The patient, evaluating physicians and family members.
  1. What are the key covariates that are used to determine treatment assignment?
  • Patient-specific factors such as patient age, prioraneurysm rupture, medical co-morbidities, social history, family history and patient preferences; aneurysm-specific factors such aneurysm size, shape and location and
  • Physician-specific factors such whether the physician making the recommendation is an endovascularspecialist, an open surgical specialist, both or neither, the years of experience of the physician and his or her practice setting.
  1.  Can we quantify the co-variants?
  • Yes
  1.  What clinically meaningful outcomes do we want to measure?
  • Mortality, aneurysmal subarachnoid hemorrhage and functional health status.
  1.  What sample size will be necessary and how long must the POD continue?  
  • These can be calculated based on available data from previous studies.

Real Answers

Given the need for a real answer to the question of optimal treatment for unruptured aneurysms, the time has come! A multicenter study collecting data using such a POD enables us to make causal inferences regarding the best treatment options for these patients would be more reliable that the causal inferences we could make from a non-generalizable RCT.9


1. Cushing, H., & Eisenhardt, L. (1938). Meningiomas, Their Classification, Regional Behaviour, Life History, and Surgical End Results. Charles C. Thomas, 785.

2. Harbaugh, K.S. & Harbaugh, R.E. (1995). Early discharge after carotid endarterectomy.  Neurosurgery 37: 219-225.

3. Harbaugh, K.S., Pikus, H.J., Shumaker, G.H., Perron, A.D., & Harbaugh, R.E. (1996). Increasing the value of carotid endarterectomy. J Neurovasc Dis 1: 40-47.

4. Harbaugh, R.E. (1997). Carotid surgery using regional anesthesia. Techniques in Neurosurgery 3: 25-33.  

5. Magnadottir, H.B., Lightdale, N., & Harbaugh, R.E. (1999). Clinical outcomes for patients at high risk who underwent carotid endarterectomy with regional anesthesia. Neurosurgery 45: 786-792.

6. Papavasiliou, A.K., Magnadottir, H.B., Gonda, T., Franz, D., & Harbaugh, R.E. (2000). Clinical outcomes after carotid endarterectomy: comparison of regional and general anesthetics. J Neurosurg 92: 291-296.

7. Harbaugh, R.E. & Patel, A. (2014). Surgical advances for extracranial carotid stenosis.  Neurosurgery 74: S83–S91.

8. Walicke, P., Abosch, A. A., Barker, F.G., Harbaugh, R., Jehi, L., Kestle, J., Koroshetz, W., Little, R., Rubin, D., Valadka, A., Wisniewski, & Chiocca, E.A. (2017). Launching Effectiveness Research to Guide Practice in Neurosurgery: Outcomes from an NINDS Workshop. Neurosurgery 80: 505-514.

9. Harbaugh, R.E. (2016). Response to editorial by Peterson EC, Heros RC. Rupture risk of unruptured aneurysms: beyond size and shape. J Neurosurg 124: 285-287.

10. Rubin, D.B. (2008). For objective causal inference, design trumps analysis. Ann Appl Stat 2: 808–840.

Leave a Reply

Steve Lownie | April 11, 2019 at 2:50 pm

Dear Dr. Harbaugh,
I’m delighted to learn you have an abiding interest in carotid endarterectomy. Carotid stenosis happens to be one of my major clinic interests too. I was interested to hear the changes your studies have made to your clinical practice. My research has been a little different (circle of Willis collateral, and plaque morphology prohibiting stenting), but had the same inspiration as yours in terms of questioning the need for monitoring.
I hope we may find a way to share data via the NPA, which we have spoken on briefly in the past. Few Canadian neurosurgery centres focus on CEA and it would be great to facilitate collaboration between U.S. and Canadian centres, as was so intrinsic to the success of NASCET. Regards, Steve