Dr. Sajeel Rehmat Khan, MD, Dr. Alexander Ivanov, MD.
University of Illinois at Chicago, Department of Neurosurgery, Chicago, IL
Journal Club Article: “The Impact of Provider Volume on the Outcomes After Surgery for Lumbar Spinal Stenosis,” by Dasenbrock et al. Neurosurgery. (June 2012).
Significance/context and importance of the study
The study is significant as it is striving to answer a question that is on the one hand straight forward and almost intuitive, namely the impact of volume on outcomes in lumbar decompression, and yet can be quite complex when applied to surgical procedures where performance measures and quality indicators could be at stake. The immediate implication here may apply to tackling learning curves, minimum number of cases needed / year to maintain competency in a desired procedure and outcomes with respect to hospital volume versus provider volume in simple vs. complex surgeries. The broader implications however are not straight forward and potentially can be extrapolated to need for regionalization / centers for excellence for complex procedures, reimbursements for procedures especially when complications are included in upcoming performance based models and federal/ local legislation re: minimum number of cases required for independent practice. This is an arena where this study or similar scientific literature using administrative data could be used by various interest groups to advance specific agendas for their own benefit, or restrict practice patterns, which may affect access to care without specifically improving overall quality and patient outcomes.
Originality of the work
Volume- outcomes estimation and correlations are not new to scientific literature and have been examined since the 70’s, 1 and recently in the Spine literature also, 2 demonstrating significant difference in complication rate in lumbar spine surgery when treated by high volume surgeons and hospitals thus demonstrating both inter operator as well as inter hospital variability.
Appropriateness of the study design or experimental approach
The study design is essentially a retrospective review of administrative data employing NIS (nationwide inpatient sample) data over 4 years using ICD codes for isolating the desired procedure (lumbar decompression and fusion) in the elective setting, identifying unique providers and hospitals for estimation of volumes and complications. The complications were also extracted using ICD 9 codes and included a broad array of neurologic as well as non neurologic complications. For the purpose of this critique, we would like to classify the complications into direct surgical complications including neurologic, unintended durotomies and surgical site infections as well as indirect surgical complications including systemic and thromboembolic complications . We propose analyzing the surgical complications separately from the systemic complications may be a better reflection of surgical finesse than classifying both groups together. It is our opinion that using administrative data for measuring volume outcome relationship can be potentially misleading as described by a recent study 3. The authors outline the main limitations of observational and administrative data sets and cite introduction of confounders and bias as the major problem areas. In this case our observations can be summed up as below, in relation to the adequacy of experimental techniques.
Adequacy of experimental techniques
The use of cumulative data (procedures/ 4 years) rather than contemporaneous data (procedures /year) may not reflect a surgeon’s volume and competency quite accurately. For example, it can be argued that a surgeon logging most of his cases in one year out of the four years may have different complication rate compared to a surgeon performing similar number of cases scattered over four years. Also administrative databases do not sub categorize experience, training (whether fellowship trained or not) and certification of a given provider and therefore can introduce another confounder during analysis. For example, it can be argued that it is quite possible that high volume surgeons in this study may more likely be fellowship trained (spine fellowship in this case) compared to low volume surgeons who might be general practice neurosurgeons and thus doing spinal surgeries less often and may be expected to have different outcomes more so because of their training rather than the volume of cases. Outcomes were measured w.r.t hospital stay, complications (both direct/ indirect surgical), hospital charges and discharge disposition. Patients’ pre operative status was based on absolute number of premorbid conditions and was adjusted accordingly in the current study. However in our opinion ASA classification would have been more accurate in assessing the preoperative status of the patient in regards to the impact on complication rates, especially systemic complications and consequently hospital stay, charges and discharge disposition.
A similar conclusion was reached in a recent article studying association between ASA classification and length of stay after orthopedic procedure,4 .One of apparent handicaps of administrative datasets is inability to calculate ASA classifications and this can introduce a potential confounder. Hospital charges tend to vary tremendously across institutions, 5 as well as states and similarly criteria for discharge disposition can vary as well, especially in the setting of patient’s preference and home situation. Additionally systemic complications can be reflective of various other confounders, for example post op cardiac and pulmonary complications may be more of a reflection of severity of preoperative comorbidities (more accurately assessed by ASA scoring) than the surgical volume and by extension surgical expertise of the involved surgeon. Analyzing surgical and systemic complications together may not reflect accurately on the technical skill of the surgeon and thus surgical volume may not be directly responsible for systemic complications. It is our observation that criteria to classify surgical volumes, and by extension demonstration of significance below a cut off, was not clearly defined. In other words, the classification could have been done by low, medium, and high categories rather than the very low, low, medium and high categories used, and possibly statistical significance might not have been achieved. It would have been useful if the criteria for establishing the volume categories were transparent. It is unclear whether they were determined apriori or after exploratory analysis? It is hard to fathom that a given surgeon performing 18 procedures / 4 years and thus classified as low volume is not expected to have significantly different mortality and complication rate than a high volume surgeon, whereas one performing 14 surgeries / 4 year is classified as very low volume and would thus be expected to have a significantly higher complication rate. Such intuitively illogical conclusions tend to fail the ‘sniff test’ and raise question about reliance wholly on statistical significance when data can potentially be categorized in different ways to achieve different statistical results.
Lumbar decompression surgeries inherently have a very low mortality rate and thus it was expected that mortality rate would not differ among groups. However, the best outcome measure of spinal surgeries is functional status, which is not reflected, in administrative data and thus prospective data/ registry data is needed to quantify this outcome measure. In our opinion, absence of functional outcomes in lumbar spine surgeries volume-outcome analysis per se significantly reduces the importance and validity of the results.
Soundness of conclusions and interpretation
The data presented support the conclusion that for simple surgeries with inherently low mortality rates and requiring low intensity post operative care ,outcomes do not differ among hospitals in various regions, of various capacities and of various roles (academic vs private setting).This may be true for simple procedures e.g lumbar decompression in this case, however with potentially high stakes surgeries with elaborate post op care requirements including well equipped ICU and multidisciplinary care requirements, one expects to find inter hospital variability and variations in mortality and complications rates therefore raising the questions of regional centers of excellence for these procedures, with resultant concentration of teaching and training opportunities. Therefore lack of interhospital variability as observed in current study for lumbar decompression and fusion may not be valid for potentially complex procedures with multi disciplinary postoperative care.
Relevance of discussion
The study discusses the impact of very low volume surgeons on volume outcomes curve and raises the question of whether a threshold exists of appropriate number of surgeries to positively impact the volume-outcomes curve. It intends to solidify the impression that even in relatively simple procedures, it is the surgeon’s expertise that is the primary determinant of outcomes. Although the conclusion may be intuitive, to reach such a conclusion based on administrative data with significant limitation both in terms of adjustment of confounders and lack of data on functional outcomes in our opinion is not well supported. The more compelling result relates to its examination of the inter hospital variabilities in outcomes for lumbar surgeries and finding that no significance exists between hospital volumes and outcomes.
Clarity of writing, strength and organization of the paper
The authors must be commended on analyzing a large data set and summarizing the findings sequentially and in a clear manner. The organization of the paper follows the accepted standards of scientific literature with sections discussing inclusion criteria, statistical methods, results and conclusion.
Economy of words
The current study summarizes the large data set appropriately and efficiently.
Relevance, accuracy and completeness of bibliography
The bibliography traces the roots of volume –outcomes analyses to the present day and as such was adequate based on our review.
Number and quality of figures, tables and illustrations
The study is quite clear in its presentation of results, and has appropriate array of tables and graphs highlighting the salient features of the study.
The most important aspect of this study is the potential implication for the future of practice of common spinal procedures, and neurosurgery as a whole. Due to inherent possibility of various confounders and inbuilt biases in administrative data sets, along with absence of functional outcomes and the unclear criteria for classification of volume categories in the current study, the results of the study do not provide strong evidence for changes in practice pattern. The unwanted and unintended consequences of using administrative datasets on volume-outcomes are potential misuse of the results of the current / similar studies to fashion debate on enhanced regulation w.r.t surgical complications, use in frivolous lawsuits, and affecting reimbursement rates in the upcoming performance based models. We opine that volume outcome analysis would best be studied in a prospective manner inclusive of functional outcomes for formulation of proper principles with respect to future practice of surgical disciplines.
Luft HS, Bunker JP, Enthoven AC. Should operations be regionalized? The empirical relation between surgical volume and mortality. N Engl J Med 1979; 301: 1364–1369.
Farjoodi P, Skolasky RL, Riley LH. The effects of hospital and surgeon volume on postoperative complications after lumbar spine surgery. Spine 2011; 36:2069–2075.
William J. French, MD; Vanessa S. Reddy, MS; Hal V. Barron, MD. Transforming Quality of Care and Improving Outcomes After Acute MI Lessons From the National Registry of Myocardial Infarction. JAMA. 2012;308(8):771-772. doi:10.1001/jama.2012.9500
Garcia AE, Bonnaig JV, Yoneda ZT, Richards JE, Ehrenfeld JM, Obremskey WT, Jahangir AA, Sethi MK.Patient Variables which may Predict Length of Stay and Hospital Costs in Elderly Patients with Hip Fracture. J Orthop Trauma. 2012 Jul 24. [Epub ahead of print]
Redelmeier DA, Bell CM, Detsky AS, Pansegrau GK. Charges for medical care at different hospitals.Arch Intern Med. 2000 May 22;160(10):1417-22.