The volume of surgery over the past several decades has increased dramatically in all parts of the world, with an estimated 234 million operations performed annually, making safe delivery of surgical care a major public health concern [10]. From the early 1900’s, it has been a tenet of the surgical profession that the careful tracking and analysis of outcomes is essential to provide safe, high-quality care [1]. A simple, low-cost metric assessing post-operative complications, capable of providing rapid feedback to the surgical teams in any setting could therefore aid clinical care and quality improvement efforts.
The concept of severity weighting used to calculate the PMI is derived from utility weighting, which is the mathematical method of assigning value weights to multidimensional outcomes states to reflect overall impact [9]. The value of severity weights used in the current study comes from a well-validated study where 50 surgical experts were asked to evaluate and score 12 clinical vignettes [11]. Thus the PMI is an index, which might be most useful in detecting trends and serve as a point of reference in the surgical field. Considering this, the PMI numbers generated in the present study should be only taken as a starting point. For example, as shown in Additional file 1: Figure S1A, the PMI could be used to follow trends in complications for any particular urological procedure at the institutional level over months or years. As demonstrated in Table 3, when using the PMI, we are no longer simply analyzing the incidence rate of complications for a particular procedure, but we are also estimating the severity score of these respective complications. Furthermore, we can analyze the expected and actual severity grades of each complication that occurs. For instance, the post-operative complication rate of a transurethral prostatectomy was 35.63%, which could be further sub-classified into severity grades and compared to the complication grades with other procedures. This can be a valuable tool in standardizing practice or research, but also can be useful in properly counseling patients of their surgical risks. For example, one can advise that the severity of complications after a transurethral prostatectomy are approximately five times higher than the severity of a complication following a ureteral catheterization (PMI 0.10 vs. PMI 0.02, respectively).
It is evident that radical cystectomy at our institution had the most frequent complication rate and that most of the morbidity related to radical cystectomy was due to Grade 4 and Grade 5 complications, as shown in Additional file 1: Figure S1A. This high complication rate was noted by other investigators [12–14]. Specifically, DeNuzio and colleagues reported 415 complications in 302 patients undergoing cystectomy and classified these complications as Clavien type I (109 patients), II (220 patients), IIIa (45 patients), IIIb (22 patients), IV (11 patients) and V (8 patients) [14]. Furthermore, ureteral catheterization and ESWL had similar PMIs, but closer analysis shows that the burden of complications of ESWL came from Grade 4 complications, making it more severe than the burden from ureteral catheterization. Another important application of the PMI is to detect trends in deterioration or improvement in surgical outcomes, particularly after the institution of corrective measures or protocols. When analyzing this data and comparing with the curves in Additional file 1: Figure S1A, which displays the percentages of complications using the unweighted severity grades by procedure, similar PMI scores show completely different severity grade distributions. For example, when comparing nephroureterectomy and transurethral prostatectomy, both procedures have similar PMI scores (0.104 vs. 0.107), however transurethral prostatectomy has a greater number of grade 1 complications. In contrast, comparison of the curves in Additional file 2: Figure S1B, which shows the burden of weighted severity grades by procedure group, we can observe that most of the severity score for the transurethral prostatectomy is derived from grade 4 complications rather than from grade 1 as observed in the unweighted plot. In our series, we realized that almost all grade 4 complications in the transurethral prostatectomy were secondary to the need of a further resection of the prostate (data not shown). It would then be possible to further examine the indications for reoperation in our case series. Furthermore, this information can be used to compare surgeons, hospitals and procedures as well as initiate studies in order to determine the causes of such complications in that particular procedure. This technique of assessment then quality improvement could ultimately enhance the level of care provided to the urologic patient.
One limitation of this method is that it still lacks absolute objectivity in rating complications. For instance, there is no way to factor in whether the high reoperation rate after transurethral prostatectomy was due to the natural disease process of prostatic hypertrophy or to inadequate gland resection. Furthermore, we applied the severity weights of these ACS NSQIP 30-day morbidity derived from our general surgery colleagues. We believe these weights should be transferrable to the urologic patient, seeing that they are based on general medical/surgical tenets, but a follow-up study will validate these findings with urologic surgeons. Next, the patient population may have a higher comorbidity or some other factor that predisposes them to have a higher post-operative risk profile. These are important factors that must be factored into such a comprehensive system, however they are beyond the scope of this project. Thus, the technique utilized by our study of calculating PMI scores has limitations that must be considered before it can be applied in any institution or in any clinical situation. In concordance with Strasberg et al., we believe that a) the simplicity of the PMI makes it an easy tool to implement but at the same time, a tool lacking the ability to perform individual risk adjustment, b) the fact that only the most serious complication in each patient is considered in order to calculate the PMI may tend to lose certain information when a patient presents with multiple complications, c) the PMI might be less useful for detecting differences across urological care providers at any point in time and d) the application of the PMI should be adequate for the majority of procedures, except for that ones with unusual complication, for example in the area of transplantation where the death of a living donor must receive a special severity weight.