SEARCH

SEARCH BY CITATION

Keywords:

  • erectile dysfunction;
  • neurovascular bundle preservation;
  • prostate cancer;
  • radical prostatectomy;
  • surgical technique

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONFLICT OF INTEREST
  8. REFERENCES
  9. Appendix

OBJECTIVES

To prospectively evaluate whether a modified surgical technique for neurovascular bundle (NVB) preservation during radical prostatectomy (RP) is associated with an improvement in erectile function (EF) recovery after RP.

PATIENTS AND METHODS

Data from patients treated before technique modification was used to create a predictive model for EF at 6 months after RP using age, date of surgery, and nerve sparing (none vs unilateral vs bilateral) as predictors for patients who received the modified technique (MT) to estimate the expected outcomes had they received the standard technique (ST), and compared these with actual outcomes. In the MT, the neurovascular bundle (NVB) is completely mobilized off the prostate from the apex to above the seminal vesicles including incision of Denonvilliers’ fascia before urethral division and mobilization of the prostate off the rectum.

RESULTS

Of 372 patients with evaluable data, 275 (74%) underwent the ST from 1 January 2001 to 31 December 2004 and 97 (26%) underwent the MT from 1 January 2005 to 30 May 2006. Sixty-five of 97 patients (67%) receiving the MT had EF recovery at 6 months, whereas the expected probability of 6-month recovery of EF, had they received the ST, was 45%. The absolute improvement in EF recovery attributable to the MT was 22% (95% confidence interval 5–40%; P = 0.013).

CONCLUSIONS

Technical modifications to NVB preservation during RP were associated with improved rates of EF recovery.


Abbreviations
NVB

neurovascular bundle

RP

radical prostatectomy

ED

erectile dysfunction

ST

standard technique

EF

erectile function

MT

modified technique

IIEF

International Index of Erectile Function

PDE-5

5-phosphodiesterase.

INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONFLICT OF INTEREST
  8. REFERENCES
  9. Appendix

Since the detailed description of the anatomical relationship between the neurovascular bundle (NVB) and capsule of the prostate [1], modifications in radical prostatectomy (RP) surgical technique have been developed to reduce the long-term morbidity of erectile dysfunction (ED) without compromising cancer control. Devising a surgical approach that minimizes treatment-related morbidity and maximally reduces time to recovery provides patients with the optimal treatment strategy.

A commonly used approach for NVB preservation is initial incision of the lateral prostatic fascia and partial mobilization of the NVB from the apical third of the gland [2,3]. This is followed by urethral transection and subsequent elevation of the apex of the prostate (usually with a Foley catheter as a tractor) to allow further lateralization of the NVB and posterior dissection of the prostate. It has been shown that early release of the NVB from the apex of the prostate before initiation of the posterior dissection reduces both the time to recovery as well as the overall likelihood of recovery of potency [4]. Nonetheless, we hypothesize that this technique (subsequently referred to as our ‘standard technique’ or ‘ST’) continues to place undue tension on the NVB resulting in traction injury and a delay in erectile function (EF) recovery, as it fails to release attachments tethering the NVB to the mid-gland and base of the prostate, Denonvilliers’ fascia, and the seminal vesicles. We modified our technique so that, before urethral division and mobilization of the prostate off the rectum, the NVB was completely mobilized off the prostate from apex to above the level of the seminal vesicles. Additionally, Denonvilliers’ fascia is incised along the length of the prostate to further lateralize the NVB, so that any manipulation of the prostate during the apical and posterior dissection results in complete protection of the NVB from traction injury. We investigated whether time to EF recovery was improved by this modified technique (MT).

PATIENTS AND METHODS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONFLICT OF INTEREST
  8. REFERENCES
  9. Appendix

Between January 2001 and May 2006, 513 consecutive men with ‘normal’ preoperative EF underwent open RP with NVB preservation (visual status 1 or 2, see below) for clinically localized prostatic adenocarcinoma. All surgical procedures were performed by one surgeon (J.A.E.). With approval of our Institutional Review Board, data were collected prospectively into our Specialized Program of Research Excellence Prostate Cancer database. All information was treated according to the Health Insurance Portability and Accountability Act and was de-identified before analysis. Patients who received prior radiation and/or chemotherapy (one patient) or neoadjuvant hormonal therapy (22) were excluded, leaving 490 patients in our sample for analysis.

All patients underwent a meticulous pelvic lymph node dissection followed by RP. The operative approach serving as our ‘ST’ has been previously described (Fig. 1) [2–5]. Modifications in our approach to NVB preservation began immediately after ligation and division of the dorsal venous complex. Depending upon preoperative and intraoperative assessment of the extent and location of disease, we used either an intrafascial or interfascial plane of dissection to release the NVB off the prostatic capsule [5]. This dissection was initiated at the apex of the prostate and continued along the posterolateral aspect to above the level of the seminal vesicles (Fig. 2). Denonvilliers’ fascia was incised by sharp dissection from the prostatic apex to the base facilitating complete mobilization and lateralization of the NVB, ensuring that any subsequent manipulation of the prostate avoids traction and potential injury to the NVB. The division of the urethra followed by the apical and posterior dissection was then completed. Patients who received the ST and MT were treated, respectively, from 1 January 2001 to 31 December 2004 and from 1 January 2005 to 30 May 2006.

image

Figure 1. ST: The NVB has been released from the apical third of the prostate. The urethra has been completed divided and the anastomotic sutures have been placed. A catheter has been placed through the urethra into the bladder. Upward traction on the catheter allows the lateral vascular pedicles to be easily isolated (A), controlled with clips (B), and divided to expose the lateral aspect of the seminal vesicle (C).

Download figure to PowerPoint

image

Figure 2. MT: The entire NVB has been dissected off the lateral aspect of the prostate from just beyond the apex to above the seminal vesicle. Denonvilliers’ fascia is sharply incised along the length of the prostate to further lateralize the NVB. Minimal traction is placed on the prostate/NVB during this dissection. The urethra is left intact until after both NVBs have been released from the prostate.

Download figure to PowerPoint

The following data were collected prospectively: preoperative clinical characteristics (PSA level, clinical stage and biopsy Gleason scores), intraoperative visual assessment of the quality of NVB preservation (rated as: 1, preserved; 2, possible damage; 3, definite damage; and 4, complete resection), estimated blood loss and surgical time, pathological features of the RP specimen (established extracapsular extension, seminal vesicle invasion and lymph node metastases, presence of tumour at the inked margin of resection, specimen Gleason score), administration of adjuvant radiation or hormonal therapy, and all imaging and laboratory data. Our follow-up protocol during the first year required serum PSA measurements every 3 months. If the results were under the upper limit of detection of the assay, the PSA level was assessed semi-annually during the second year and annually thereafter.

At each visit, outcomes regarding EF were evaluated by the treating physician using patient-reported questionnaires assessing for the quality and frequency of sexual activity and graded on a 5-point rigidity scale (Table 1). We previously published data showing good agreement between this 5-point scale and the International Index of Erectile Function (IIEF) for defining functional recovery of EF after RP [6]. Additionally, the type and start date of medical treatment for ED was recorded. Any relevant EF information obtained through clinical protocols and patient surveys were uploaded into our database system by research assistants.

Table 1.  5-point rigidity scale vs IIEF
Potency levelLevel definitionMean IIEF (95% CI)
1Normal, full erections21.2 (17.5–25)
2Full, but diminished erections satisfactory for sexual activity19.6 (17–22.5)
3Partial erections occasionally satisfactory for sexual activity14.8 (12–17.5)
4Partial erections unsatisfactory for sexual activity 6 (4.3–8)
5No erection 2.7 (1–4)

All patients were encouraged to use oral 5-phosphodiesterase (PDE-5) inhibitors after RP once the Foley catheter was removed [7,8]. EF recovery after RP was recorded if an erectile rigidity score of 1 or 2 was achieved with or without the use of PDE-5 inhibitors. Patients with erectile rigidity scores of 3–5 or those requiring the assistance of vacuum devices or injection therapy to achieve erections suitable for penetration were considered to have ED. Because the development of neuropraxia and subsequent postoperative ED is transient and proportional to the degree of traction injury sustained at RP, the evaluation of EF recovery was centred on short-term outcomes at 6 months after RP to assess for differences in the rate of EF recovery.

The EF data were subject to interval censoring: e.g. consider a patient seen in clinic at 3 months and noted to have ED, who then returns at 9 months and reports full recovery of EF. Although this man regained EF somewhere between 3 and 9 months, in a traditional Kaplan–Meier survival type analysis, such a patient would be counted as having an ‘event’ at 9 months. To avoid interval censoring in this series, we first converted the survival-time data to binary data for the outcome of 6-month EF. For patients evaluated for EF recovery between 4.5 and 7.5 months after surgery, we used the evaluation closest to 6 months (314 patients). For patients not evaluated between 4.5 and 7.5 months after RP, we included those with an evaluation of <4.5 months after RP and an evaluation of >7.5 months after RP, if the EF status was the same on both occasions (58); otherwise, patients were excluded from analysis (118). Of the 118 patients excluded, all were evaluated for EF recovery on at least one occasion after RP; 73 (62%) were not evaluated after 4.5 months; 21 (18%) were not evaluated before 7.5 months; and 24 (20%) had ED before 4.5–7.5 months but recovered EF after that.

Our aim was to evaluate the association between surgical technique (ST vs MT) and EF recovery after RP, controlling for age at RP and nerve-sparing status (none vs unilateral vs bilateral). This analysis is complicated by changes in treatment of ED after RP and improved surgical EF outcomes with greater experience of the treating surgeon (the ‘learning curve’). Our approach was therefore to compare the actual outcome for patients who received the MT with their expected outcome had they been treated using the ST, which was determined by a predictive model created with data from patients who received the ST. We fitted a logistic regression model using only patients who received the ST; the outcome of this model was 6-month EF recovery and the predictors were date of surgery, age at RP, and nerve-sparing status. Inclusion of the date of surgery in this model controlled for the surgical learning curve and secular changes in the treatment of ED. From the model, we obtained the predicted probability of 6-month EF recovery for patients who received the MT. The predicted probability was compared with the actual EF recovery status at 6 months using a statistic, denoted by S, which is the mean difference between actual EF recovery status at 6 months and the predicted probability from the model among patients who received the MT. This statistic, S, can be interpreted as the estimated improvement in EF recovery attributable to the use of the MT. To obtain a CI for S, we first define Y as the indicator variable for EF recovery (1, EF recovery; 0, no EF recovery); ρ as the predicted probability of 6-month EF recovery from the logistic regression model described above; n as the number of patients receiving the MT. We can then formally define S as inline image. To obtain a central estimate and se for S we used a bootstrapping method (Appendix).

To check our results, we used a previously developed method of comparing predictions with results [9]. An analogous method was used to illustrate our results graphically: the logistic regression was run separately for ST and MT patients and the two lines plotted on one graph.

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONFLICT OF INTEREST
  8. REFERENCES
  9. Appendix

In all, 118 patients were excluded because of unknown EF recovery status at 6 months after RP, thus 372 men in our cohort provided evaluable data: 275 patients (74%) underwent the ST and 97 (26%) underwent the MT (Table 2). There were no important differences between the groups regarding age distribution, operative time, estimated blood loss, preoperative features of the prostate cancer, or pathological features of the RP specimen (Table 2). The positive surgical margin rates were 9% and 10%, respectively (P = 0.8, Fisher’s exact test).

Table 2.  Patient characteristics stratified by inclusion in statistical analysis and by surgical technique (ST or MT)
VariableExcluded from analysis (n = 118)Included in analysis (n = 372)
STMTSTMT
N 100  18 275 97
Median (interquartile range):
 Age at surgery, years  57 (53, 61)  55 (50, 62)  57 (53, 61) 59 (52, 63)
 Preoperative PSA level, ng/mL   5.28 (4.28, 7.37)   4.27 (3.29, 7.78)   5.1 (4.18, 6.9)  4.84 (3.38, 7.22)
 Operating room time, min (n = 485) 180 (150, 180) 165 (150, 180) 180 (150, 180)165 (150, 180)
 Estimated blood loss, mL (n = 485) 1100 (750, 1500)1200 (900, 1200)1000 (700, 1500)900 (700, 1225)
Frequency, n (%)
 Clinical stage (n = 480)
  T1  67 (67)  12 (67) 178 (65) 64 (66)
  ≥T2  33 (33)   4 (25)  95 (35) 27 (30)
 Biopsy Gleason grade (n = 453)
  ≤6  60 (60)   7 (39) 179 (65) 56 (58)
  7  26 (26)   7 (39)  63 (23) 30 (31)
  ≥8   5 (5)   2 (11)  10 (4)  8 (8)
 Pathology Gleason grade (n = 484)
  ≤6  47 (47)   7 (39) 144 (52) 38 (39)
  7  50 (50)   9 (50)  117 (43) 54 (56)
  ≥8   2 (2)   1 (6)   11 (4)  4 (4)
 Extracapsular extension (n = 486)  10 (10)   6 (33)  65 (24) 29 (30)
 Seminal vesical invasion (n = 485)   2 (2)   1 (6)   11 (4)  6 (6)
 Lymph node involvement (n = 471)   3 (3)   1 (6)   2 (1)  3 (3)
 Positive surgical margins   9 (9)   3 (17)  26 (9) 10 (10)
 Reasons for exclusion:  
  Not evaluated after 4.5 months  61 (61)  12 (67)
  Not evaluated before 7.5 months  19 (19)   2 (11)
  ED before 4.5 months, potent after 7.5 months  20 (20)   4 (22)

In total, 176 (47%) men had functional erections at 6 months after RP. The 6-month predicted probability of EF recovery, with adjustment for age and nerve-sparing status, is shown in Fig. 3 separately by surgical technique. As expected, the EF recovery rates improved with time, irrespective of surgical technique. Notably, improvement in EF recovery rates spiked immediately after the surgeon commenced the MT. To formally test for an improvement in EF based on surgical technique, we compared the observed outcome in men who received the MT against their expected outcome, had they received the ST. In all, 111 of 275 patients (40%) who received the ST and 65 of 97 patients (67%) receiving the MT had EF recovery at 6 months after RP. The expected probability of 6-month EF recovery in patients undergoing the modified surgery, had they received the ST, was 45%. Therefore, the absolute improvement in EF apparently attributable to the new surgical technique is 22% (95% CI 5–40%; P = 0.013). Use of the Heller method [9] gave extremely similar results (absolute improvement of 22%; 95% CI 4–40%; P = 0.015).

image

Figure 3. Predicted probability of potency at 6 months after RP by surgical technique, with adjustment for age and nerve sparing status, with 95% CIs. Years 2001–2004 ST; years 2005–2006: MT. The 6-month probability of potency, adjusted for age and nerve-sparing, for each year of surgery is denoted by an ‘x’.

Download figure to PowerPoint

The results for the subset of 269 men who received bilateral NVB preservation (visual status 1 bilaterally) were very similar: the expected probability of 6-month EF recovery for the 53 men who received the MT, had they received the ST, was 57%; 42 of those men (79%) had EF recovery at 6 months, resulting in an absolute improvement of 22% (95% CI 2–41%; P = 0.03).

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONFLICT OF INTEREST
  8. REFERENCES
  9. Appendix

Surgical treatment of clinically localized prostate cancer has focused on decreasing inherent morbidity. Historically, ED had been universal in patients undergoing RP. However, through anatomical studies performed by Walsh and Donker [1], identification of the course and relationship of the NVB relative to the prostatic capsule has allowed for intraoperative modifications facilitating recovery of EF [10].

Increased surgeon experience and surgical modifications have been shown to improve EF outcomes [11]. In addition to preoperative EF, NVB preservation, and patient age, fine details in surgical technique have proven to be independent predictors for recovery of EF after RP [12]. Through retrospective review of intraoperative videotapes, Walsh et al. [13] were able to identify four modifications that appeared to correlate with recovery of EF. More recently, they reported improvements in 18-month EF recovery rates among 86% of selected patients with and without the use of PDE-5 inhibitors [14]. Similarly, use of intraoperative video documentation and pathological review at our own institution facilitated identification of technical aspects during laparoscopic RP that contributed to reduced positive surgical margin rates [15].

The concept of early release of the NVB is not new [2–4]. Chuang et al. [4] reported their experience in early release after anterior division of the urethra before the posterior dissection by showing improved rates of overall EF recovery with both use of optical magnification and early release of the NVB. However, it remains uncertain whether traction placed on the NVB is completely circumvented when release is confined to the apical third of the prostate. Based upon surgeon experience and intraoperative observations, NVB preservation techniques were modified in the present study. In an effort to minimize traction, we now completely mobilize the NVB off the prostate from the apex to above the seminal vesicles and further lateralize the NVB by incising Denonvilliers’ fascia along the entire length of the prostate before dividing the urethra. We hypothesized this would reduce the degree of neuropraxia encountered postoperatively that might contribute to a delayed recovery of EF after RP. Because neuropraxia is transient and proportional to the degree of traction injury, we assessed for differences in short-term outcomes for postoperative EF recovery. In the present study, patients undergoing the MT had 6-month EF recovery rates that were 22% higher than what would have been expected had they received the ST. Importantly, we saw no evidence that the MT compromised cancer control, as the rate of positive surgical margins was similar for the groups.

Although further evaluation of EF recovery rates at 12, 18, and 24 months will be required to assess the effect our change in technique has on overall EF recovery, evidence suggests that the effect of delayed EF recovery beyond 6 months results in corporal fibrosis predisposing to venous leak and secondary ED despite resolution of the postoperative neuropraxia [16]. With earlier recovery of EF, longer-term recovery rates might be improved. The 6-month rate of EF recovery of 67% (79% in those undergoing bilateral neurovascular bundle preservation) is comparable to other contemporary published series. Walsh et al. [14] reported 54% recovery rate at 6 months in 64 men undergoing anatomic (bilateral NVB preservation) RP with and without the use of PDE-5 inhibitors. The present 6-month EF results compare favorably to those of Menon et al. [17] who recently published 12- and 48-month EF recovery results in men undergoing their modified technique of robotic RP. In men in whom a bilateral preservation of the ‘veil of Aphrodite’ was achieved (only 33% of their cohort), 70% had return of EF at 12 months after RP.

The functional outcomes in the present study are subject to several limitations. Variations in either the patient’s perceived or actual baseline EF, individual surgeon assessment of the quality of NVB preservation, and patient compliance with and/or physician recommendations for the use of postoperative PDE-5 inhibitors all potentially influence our ability to interpret improvements in the rate of EF recovery attributable only to changes in technique. Additionally, the rigidity assessment scale used to assign the degree of EF recovery requires further validation at other centres; although, we have previously published data showing good agreement between levels 1 and 2 on the 5-point scale and functional outcomes as defined by the EF domain score of the IIEF [6]. In our previous publication, the mean EF domain score equating to a level 2 erection was 19.6 (95% CI 17–22.5). This translates to a per question mean of 3.3 indicating the ability to have sexual intercourse between ‘half the time’ (3) and ‘most of the time’ (4). The IIEF is validated for all forms of sexual activity, therefore we would contend so too is the 5-point erection scale for such activity. Furthermore, expected improvement in oncological and functional outcomes related to both the individual surgeon’s progression along a learning curve with increased experience and the influence of interval censoring on data interpretation during the postoperative period also potentially affect results. Nonetheless, the effect of technique modification on time to recovery after controlling for secular changes appears to remain significant.

In conclusion, the technical modifications made in our approach to NVB preservation during RP were associated with a shorter time to return of EF. Whether a causal association exists between change in technique and outcome requires further evaluation. Additionally, a longer follow-up will be required to determine the impact our MT has on the quality of erections, overall potency rates, and oncological outcome.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONFLICT OF INTEREST
  8. REFERENCES
  9. Appendix
  • 1
    Walsh PC, Donker PJ. Impotence following radical prostatectomy: insight into etiology and prevention. J Urol 1982; 128: 4927
  • 2
    Eastham JASP. Treatment of early stage prostate cancer: radical prostatectomy for clinical stage T1 and T2 prostate cancer. In VogelzangNJ, ScardinoPT, ShipleyWU, CoffeyDS eds, Genitourinary Oncology, 2nd edn. Philadelphia: Lippincott, Williams & Wilkins, 2000: 72238
  • 3
    Montorsi F, Salonia A, Suardi N et al. Improving the preservation of the urethral sphincter and neurovascular bundles during open radical retropubic prostatectomy. Eur Urol 2005; 48: 93845
  • 4
    Chuang MS, O’Connor RC, Laven BA, Orvieto MA, Brendler CB. Early release of the neurovascular bundles and optical loupe magnification lead to improved and earlier return of potency following radical retropubic prostatectomy. J Urol 2005; 173: 5379
  • 5
    Myers RP, Villers A. Anatomic considerations in radical prostatectomy. In KirbyRS, PartinAW, FeneleyM, ParsonsJK eds, Prostate Cancer: Surgical Principles and Practice, Chapt. 71. Abingdon: Taylor and Francis, 2006: 70113
  • 6
    Secin FP, Koppie TM, Scardino PT et al. Bilateral cavernous nerve interposition grafting during radical retropubic prostatectomy: Memorial Sloan-Kettering Cancer Center experience. J Urol 2007; 177: 6648
  • 7
    Zippe CD, Kedia AW, Kedia K, Nelson DR, Agarwal A. Treatment of erectile dysfunction after radical prostatectomy with sildenafil citrate (Viagra). Urology 1998; 52: 9636
  • 8
    Schwartz EJ, Wong P, Graydon RJ. Sildenafil preserves intracorporeal smooth muscle after radical retropubic prostatectomy. J Urol 2004; 171: 7714
  • 9
    Heller G, Kattan MW, Scher HI. Improving the decision to pursue a phase 3 clinical trial by adjusting for patient-specific factors in evaluating phase 2 treatment efficacy data. Med Decis Making 2007; 27: 3806
  • 10
    Walsh PC. Anatomic radical prostatectomy: evolution of the surgical technique. J Urol 1998; 160: 241824
  • 11
    Kundu SD, Roehl KA, Eggener SE, Antenor JA, Han M, Catalona WJ. Potency, continence and complications in 3477 consecutive radical retropubic prostatectomies. J Urol 2004; 172: 222731
  • 12
    Rabbani F, Stapleton AM, Kattan MW, Wheeler TM, Scardino PT. Factors predicting recovery of erections after radical prostatectomy. J Urol 2000; 164: 192934
  • 13
    Walsh PC, Marschke P, Ricker D, Burnett AL. Use of intraoperative video documentation to improve sexual function after radical retropubic prostatectomy. Urology 2000; 55: 627
  • 14
    Walsh PC, Marschke P, Ricker D, Burnett AL. Patient-reported urinary continence and sexual function after anatomic radical prostatectomy. Urology 2000; 55: 5861
  • 15
    Touijer K, Kuroiwa K, Saranchuk JW et al. Quality improvement in laparoscopic radical prostatectomy for pT2 prostate cancer: impact of video documentation review on positive surgical margin. J Urol 2005; 173: 7658
  • 16
    Nandipati KC, Raina R, Agarwal A, Zippe CD. Erectile dysfunction following radical retropubic prostatectomy: epidemiology, pathophysiology and pharmacological management. Drugs Aging 2006; 23: 10117
  • 17
    Menon M, Shrivastava A, Kaul S et al. Vattikuti Institute prostatectomy: contemporary technique and analysis of results. Eur Urol 2007; 51: 64858

Appendix

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONFLICT OF INTEREST
  8. REFERENCES
  9. Appendix

Bootstrap method for computing the central estimate and standard error (se) of S

  • 1
    Sample with replacement from the patients receiving the ST
  • 2
    Fit the logistic regression model using the sample in (1).
  • 3
    Sample with replacement from the patients receiving the MT
  • 4
    Apply the fitted model in (2) to the sample in (3) to obtain the predicted probability of 6-month EF recovery for this sample
  • 5
    Compute S
  • 6
    Repeat steps 1–5 a total of 10 000 times
  • 7
    Calculate statistics:
  • a.
     The mean value of S over all replications is the central estimate of treatment effect.
  • b.
     The se of S, denoted by se (S), is given by the sd of S over all replications.
  • c.
     A 95% CI is given by S ± 1.96 ×se (S).
  • d.
     The test statistic S/se (S) can be related to a standard normal distribution to obtain the P value associated with null hypothesis of no improvement.