Answers are grounded in this paper and cite their sources.
To the Editor: I read with great interest the article by Khan et al titled “Surgical Outcomes for Minimally Invasive vs Open Transforaminal Lumbar Interbody Fusion: An Updated Systematic Review and Meta-analysis.”1 The authors referred to 2 previous meta-analyses2,3 that reported no difference in complication rates between MI and standard open transforaminal lumbar interbody fusion (TLIF) procedures. They then state that this study, “larger and more current,” shows “a significantly lower rate of complications in minimally invasive transforaminal lumbar interbody fusion (MI-TLIF) patients.” There are several points, however, that undermine that assertion. First, the authors failed to note that there is considerable heterogeneity in what might be considered a complication in the studies they included in the meta-analysis. The authors' criteria for complication was that it was a “direct result of the intervention.” The problem is that I do not think all the studies recorded the same complications. The included article summary table had one study reporting “benzodiazepine withdrawal”4 as a complication in an open TLIF. Another reported wound fat liquefaction.5 Yet another included adjacent segment degeneration.6 One reported “overlong screws.”7 The problem here is that these complications are not really relevant to studying the superiority of one technique compared with another. At the same time, these “outlier” complications were likely not even recorded by many of the studies, yet they are included in the meta-analysis. The authors should have just defined a set of core complications (infection, radiculitis, cerebrospinal fluid leak, malpositioned screws, etc) that are relevant to their clinical question and then noted whether the studies reported them or not. Second, the authors did not detail how they arrived at the distinction between complications and complication events. Nevertheless, I was not impressed with how they handled it. Reading the Sulaiman and Singh8 article, for example, which was included in their analysis, I found that 1 open TLIF patient had a wound infection that required 3 separate incision and drainage (I+D) procedures to treat. The meta-analysis considered these 4 “complications” (1 for the infection, 3 for each I+D because of that infection) even though this was 1 postoperative patient and the Sulaiman article itself treated this as only 1 complication. I confess I did not peruse every other article to the same rigor as the authors—too expensive nowadays. But I am fairly confident that not every infection reported also included the separate treatments such as I+D procedures that could be recorded in a similar fashion. Plus, this definition defies common sense—that is, these 4 “complication” events were not independent occurrences. The authors noted in their results that there were 25 studies from which they gathered complications. Figure 11 in the article1 actually lists 27, of which there were 4 in which the risk ratio seemed to favor MI-TLIF. Three of these were retrospective (including the Sulaiman study). The Sulaiman study had a complication event rate of 7% (4/57) in MI-TLIF and 45% (5/11) in the open TLIF. Now, if one treats the open TLIF patient as 1 infection, the rate would be only 18% (2/11) instead, which would shift it toward favoring open TLIF. I think the way the authors represent complications in this study may introduce a bias here that they did not mention. Third, the only prospective study, by Wong et al,9 favoring MI-TLIF also happened to be the most heavily weighted study in the analysis. With their method of complication reporting, the open-TLIF cohort of 54 patients had 49 events, resulting in a complication event rate of 91% (49/54). The number of infections alone (19 reported) would result in an infection rate of 35% (19/54). Is this what is seen in reality? Several studies quote TLIF surgical site infection (SSI) rates of 5% to 8%.10,11 Now, it is possible that the infections in this study may have been counted in a manner similar to the Sulaiman study. I did not have access to the Wong article to confirm that. However, I think it is irrelevant. The entire assertion of the meta-analysis favoring MI-TLIF for lower complication rates essentially rests on the Wong study alone simply because it is the largest cohort and it heavily favored MI-TLIF. At the same time, Figure 11 shows that the Wong study had >3 times the complication events in the open group (n = 49) compared with the next largest number of complications (n = 14),12 despite doing essentially the same number of open procedures. If you treat this study as an outlier, then I don't think the complication rates between MI and open were any different. Finally, from the analysis presented and the conclusion that MI-TLIF had lower complication rates, the authors were quick to jump to the conclusion that because their study included more current data, it likely demonstrates the decreased complications associated with spine surgeons progressing along the learning curve. Looking at Figure 11, I have to disagree with this assertion. My reasoning is that the authors neglected to mention that there was a large disparity in the size of the MI vs open cohorts in every study in which the risk ratio favors MI-TLIF. The Sulaiman and Singh 8 study did 57 MI-TLIFs vs 11 open; the Wong et al9 study, 144 vs 54. I found 6 studies4,8,9,13-15 in which the MI cohort was at least 1.5 times larger than the open cohort; 4 had a risk ratio that favored MI-TLIF, and 2 showed no preference. If we separate these 6 “unequal cohort” studies (Group A) from the remaining 21 “more equal cohort” studies (Group B) in Figure 11, and add up the events again, we get the following (Table 1).TABLE 1: Comparison of Procedures and Complications in Groups With Unequal vs Equal CohortsSome interesting observations here. First, note that Group A does about 2.5 (356/141) times more MI-TLIF procedures than open ones. Group B (785/758) is about at parity. Defining the complication rate as the number of events divided by the total procedures, we get the following (Table 2).TABLE 2: Comparison of Complication Rates in Groups With Unequal vs Equal CohortsWhen the data is presented in this fashion, it would seem as if surgeons who do a much larger number of MI-TLIF cases have a higher complication rate with their open-TLIF cases compared with surgeons who split their caseload more evenly. I confess that I did not do a full statistical analysis here. Yet the difference in open complication rates seems pretty large. From this, one could argue that the complication disparity is occurring not because surgeons are becoming more proficient in MI techniques. Instead, as their MI caseload becomes significantly larger than their open one, surgeons may be having difficulty maintaining proficiency in both. Surgeons may be progressing along the MI learning curve, but they could also just be regressing along the open one. On a technical note, the original article's summary table of studies used in the analysis did not have citations to the publications from which these studies were retrieved. I found that the article's references did not include at least 10 publications that were used in this meta-analysis. I have referenced some of these myself.8,13,15 Disclosure The author has no personal, financial, or institutional interest in any of the drugs, materials, or devices described in this article.
No takes yet. Share an insight, caveat, or question.
Sandeep S. Bhangoo (2016) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: