Peritonitis is a major cause of morbidity and mortality

Proceedings of the First Asian Chapter Meeting ISPD December 13 15, 2002, Hong Kong Peritoneal Dialysis International, Vol. 23 (2003), Supplement 2 0896-8608/03 $3.00 +.00 Copyright 2003 International Society for Peritoneal Dialysis Printed in Canada. All rights reserved. A RANDOMIZED, MULTICENTER, OPEN-LABEL TRIAL TO ESTABLISH THERAPEUTIC EQUIVALENCE BETWEEN THE CAREX AND ULTRA DISCONNECT SYSTEMS IN PATIENTS ON CONTINUOUS AMBULATORY PERITONEAL DIALYSIS Loke-Meng Ong, 1 Teck-Onn Lim, 2 Lai-Seong Hooi, 3 Zaki Morad, 2 Poh-Choo Tan, 2 Hin-Seng Wong, 2 Yam-Ngo Lim, 4 Rozina Ghazalli, 1 Chwee-Choon Tan, 5 Wan Shaariah, 6 and Boon-Seng Liew 7 Department of Medicine, 1 Penang Hospital; Department of Nephrology, 2 Institute of Urology and Nephrology, Kuala Lumpur Hospital; Haemodialysis Unit, 3 Sultanah Aminah Hospital, Johor Bahru; Department of Paediatrics, 4 Institute of Paediatrics, Kuala Lumpur Hospital; Haemodialysis Unit, 5 Tengku Ampuan Rahimah Hospital, Kelang; Department of Medicine, 6 Seremban Hospital; and Department of Medicine, 7 Queen Elizabeth Hospital, Sabah, Malaysia Objective: In the present study, we undertook to establish therapeutic equivalence with respect to peritonitis and technique failure between the Carex disconnect system (B. Braun Carex, Mirandola, Italy) and the standard Ultra system (Baxter Healthcare, Tokyo, Japan) in patients on continuous ambulatory peritoneal dialysis (CAPD). Design: This multicenter, parallel group, randomized controlled trial involved 363 prevalent CAPD patients from 8 centers. The primary endpoint was peritonitis rate; secondary endpoints were technique failure and technical problems encountered. The duration of the evaluation was 1 year. Results: The risk of peritonitis on Carex varied between the centers. We found a significant treatment-center interaction effect (likelihood ratio test: p = 0.03). The incidence rate ratio (IRR) of peritonitis on Carex as compared with Ultra ranged from 0.4 to 7.2. In two centers, Carex was inferior to Ultra with regard to peritonitis; but, in five centers, the results were inconclusive. Equivalence was not demonstrated in any center. The overall rate of peritonitis in the Carex group was twice that in the Ultra group [IRR: 2.18; 95% confidence interval (CI): 1.51 to 3.14]. Technique failure and technical problems were more common with the Carex system. Technique failure rate at 1 year was 44% in the Carex group and 22% in the Ultra group. Conclusions: Equivalence between the Carex disconnect system and the Ultra disconnect system could not be demonstrated. The risk of peritonitis on Carex varied significantly between centers. Perit Dial Int 2003; 23(S2):S139 S143 www.pdiconnect.com KEY WORDS: Continuous ambulatory peritoneal dialysis; peritonitis; Carex; Ultra; disconnect systems. Correspondence to: T.O. Lim, Clinical Research Centre, Hospital Kuala Lumpur, Jalan Pahang, Kuala Lumpur 50586 Malaysia. limto@crc.gov.my Peritonitis is a major cause of morbidity and mortality in continuous ambulatory peritoneal dialysis (CAPD) patients and the most common cause of dropout from CAPD. Advances in connectology have resulted in reduced rates of peritonitis. The Y disconnect system was shown to be superior to the singlebag system with regard to peritonitis rate (1 4). Before the present study, most patients under Ministry of Health (MOH) care were using the Ultra disconnect system (Baxter Healthcare, Tokyo, Japan). In 1999, another Y disconnect system, Carex (B. Braun Carex, Mirandola, Italy), was introduced. For contractual reasons, we were required to convert our patients to the Carex system. During the conversion, we took the opportunity to evaluate the Carex system in comparison with Ultra. PATIENTS AND METHODS Our multicenter, parallel-group, randomized controlled trial enrolled 363 prevalent CAPD patients who were using the Ultra system at 8 MOH centers. We excluded children aged less than 12 years and older patients who had a terminal illness and life expectancy of less than 6 months, or those who were expected to change the modality of renal replacement therapy over the next 3 months, or those who required cycler assistance. Table 1 shows the distribution of subjects by center. Because of the nature of the product, blinding was not possible. The primary endpoint was peritonitis rate. Secondary endpoints were technique failure and technical problems with the systems. Peritonitis was defined as presence of at least two of the following: (A) abdominal pain or tenderness; (B) presence of white blood cells in the peritoneal effluent in excess of 100 cells per milliliter, composed of at least 50% polymorphs; (C) positive culture. Technique failure S139

PROCEEDINGS OF THE FIRST ASIAN CHAPTER MEETING ISPD DECEMBER 2003 VOL. 23, SUPPL 2 PDI S140 TABLE 1 Randomized Distribution of Patients by Center Center Carex (n) Ultra (n) Distribution [n (%)] 1 6 8 14 (4) 2 8 7 15 (4) 3 35 34 69 (19) 4 16 12 28 (8) 5 19 17 36 (10) 6 70 68 138 (38) 7 26 31 57 (16) 8 4 2 6 (2) Total 184 179 363 (100) was defined as transfer to hemodialysis, crossover to comparator, or death. The randomization list was generated separately for each participating center, using the random permuted block method with randomly varying block size (5). Subjects randomized to Ultra remained on the pre-existing Ultra system. At conversion, subjects randomized to Carex required a change in transfer set and training in the use of the new system. For the Ultra group, the duration of the evaluation was 1 year, commencing May 1, 1999. Conversion to Carex commenced May 3, 1999, and was completed in 5.5 months. Following conversion, a wash-in period of 1 month was allowed for patients randomized to the Carex system on the assumption that the conversion process exposes patients to an increased risk of peritonitis. Thus the observation period for Carex was 13 months. All reported episodes of peritonitis were subject to verification by two nephrologists who were blinded to patient s treatment assignment and were independent of the center reporting the peritonitis. Only verified peritonitis episodes were included in analysis. In the Carex group, peritonitis episodes occurring during the wash-in period were excluded from the analysis. The rate of peritonitis on Carex as compared with that on Ultra is expressed as an incidence rate ratio (IRR). Likelihood ratio was used to test for differences in the peritonitis rate between centers. Treatmentcenter interactions were tested using the Poisson model, which was significant. Hence, center-specific IRR was estimated using a fixed-effect Poisson model, including an interaction term. The combined treatment effect was estimated by taking the average of the effects of individual centers, weighted by the inverse of the variance between individual centers. The Stata statistical software package (Stata Corporation, College Station, TX, U.S.A.), release 5.0, was used for analysis. The margin of equivalence between Carex and Ultra was accepted as an IRR of 0.8 1.2. To conclude therapeutic equivalence, the upper and lower limits of the 95% confidence interval (CI) of the IRR of peritonitis on Carex as compared with Ultra had to lie entirely within those limits in both the per-protocol (PP) and intention-to-treat (ITT) analyses (6). The sample size required was based on an expected peritonitis rate of 0.05 episodes per patient month and the pre-determined margin of equivalence. For a power of 0.8 and an α of 0.05 (one-sided), the required sample size was 206 patients per arm. RESULTS Figure 1 shows the disposition of patients during the trial. Table 1 shows the distribution of patients by center, and Table 2, the baseline characteristics of the patients. No significant differences were observed in baseline characteristics between the two groups. Figure 1 Trial profile.

PDI DECEMBER 2003 VOL. 23, SUPPL 2 PROCEEDINGS OF THE FIRST ASIAN CHAPTER MEETING ISPD TABLE 2 Patient Demographics and Baseline Characteristics Carex (n=184) Ultra (n=179) Age (years) Median 47 47 Range (12 80) (14 78) Men [n (%)] 76 (41) 74 (41) Primary renal disease [n (%)] Unknown 46 (25) 44 (25) Diabetic nephropathy 50 (27) 42 (23) Glomerulonephritis 38 (21) 48 (27) Obstructive nephropathy or calculi 21 (11) 19 (11) Others 29 (16) 26 (14) Comorbidity [n (%)] Diabetes 52 (28) 44 (25) Impaired vision 17 (9) 11 (6) Ischemic heart disease 28 (15) 22 (12) Cerebrovascular disease 7 (4) 1 (1) Duration on CAPD (months) Median 25.5 25.2 Range (0.72 115.8) (0.53 92.2) Previous peritonitis rate Episodes/patient month 0.037 0.031 95% CI (0.033 to 0.043) (0.027 to 0.037) Peritoneal dialysis regimen [n (%)] Conventional CAPD 176 (96) 171 (96) DAPD 8 (4) 8 (4) Exchanges 4 daily 182 (99) 174 (97) CAPD assistance required 45 (25) 33 (18) Lab tests results [median (interquartile range)] Hemoglobin (g/dl) 9.2 (2.7) 9.1 (2.5) Albumin (g/l) 35 (7) 35 (8) Dialysis CCr (L/week) 50 (12) 50 (14) Renal CCr (L/week) 2 (7) 1 (5) Dialysis Kt/V 2.0 (0.7) 1.9 (0.7) Renal Kt/V 0.03 (0.22) 0.02 (0.22) CAPD = continuous ambulatory peritoneal dialysis; CI = confidence interval; DAPD =daytime ambulatory peritoneal dialysis (dry night); CCr = creatinine clearance. PERITONITIS We recorded 185 episodes of peritonitis, of which 148 (80%) were verified. The proportion of peritonitis verified in the Carex and Ultra groups was 82% and 76% respectively. In the Carex group, 100 episodes of verified peritonitis occurred in 952 patient months (9.5 patient months per episode). The Ultra group had 48 episodes of peritonitis in 1014 patient months (21.1 patient months per episode). The peritonitis rate varied by center in both groups. In the Carex group, the peritonitis rate ranged from 4.9 patient months to 36.1 patient months per episode. In the Ultra group, the peritonitis rate ranged from 12.2 patient months to 34.6 patient months per episode. One center (center 8) recorded no peritonitis at baseline or during the study, and we excluded that center from the analysis. Significant treatment-center interaction was found (likelihood ratio test: p = 0.03). Equivalence between Carex and Ultra with regard to peritonitis rate could not be demonstrated in any center. In two centers (centers 6 and 7), Carex was inferior to Ultra. In those centers, the lower limit of the 95% CI was greater than 1.2. Center 6 recorded an IRR of 2.9 (95% CI: 1.5 to 5.7), and center 7 recorded an IRR of 7.2 (95% CI: 2.8 to 18.5) by PP analysis (Figure 2). In the other five centers, results were inconclusive. The IRRs of peritonitis on Carex as compared with those on Ultra were 0.4 (95% CI: 0 to 4.3), 0.5 (95% CI: 0.1 to 5.6), 1 (95% CI: 0.5 to 2.1), 1.7 (95% CI: 0.5 to 5.8), and 2.5 (95% CI: 1 to 6.4) for centers 1, 2, 3, 4, and 5 respectively (Figure 2). Results recorded using the ITT approach were similar. S141

PROCEEDINGS OF THE FIRST ASIAN CHAPTER MEETING ISPD DECEMBER 2003 VOL. 23, SUPPL 2 PDI TABLE 3 Technical Problems by Disconnect System Carex (n=179) Ultra (n=176) Figure 2 Incidence rate ratio (IRR) of peritonitis on the Carex disconnect system as compared with the Ultra disconnect system, by center. To provide an overall summary of the IRR of peritonitis, we used the fixed Poisson model to combine the estimates from individual centers. The combined IRR of peritonitis on Carex as compared with Ultra was 2.18 (95% CI: 1.51 to 3.14) by PP analysis (Figure 2) and 2.1 (95% CI: 1.46 to 3.02) by ITT analysis. SECONDARY ENDPOINTS Technical Difficulties: During the conversion period, 49% of patients reported difficulties. The most common problems encountered were difficulties in closing the clip clamps (32%) and in using the Braunoderm (B. Braun Carex, Mirandola, Italy) spray (24%). Technical problems were reported more frequently with the Carex system (Table 3). In the Carex group, difficulties were more commonly reported in the initial phase of the evaluation. In the first 3 months, 49% of patients had difficulty, as compared with 38% of patients in the second 3 months, and 24% in last 6 months. Technique Failure: The technique failure rates at 3 months, 6 months, and 12 months for Carex were 11%, 21%, and 44%. For Ultra, the comparable rates were 6%, 12%, and 22%. The differences in technique failure rate occurred because of crossover from Carex to Ultra. Of the 36 patients who crossed over, 22 (61%) did so as a result of non resolving peritonitis. The other reasons for crossover were patient preference (23%), inability to use the Braunoderm spray (9%), unavailability of low-calcium dialysate with the Carex system (6%), and presence of fungal growth in the Tenckhoff catheter (3%). At 1 year, the death rate (13% on Carex, 15% on Ultra) and the rate of transfer to hemodialysis (10% on Carex, 7% on Ultra) were not significantly different between the two groups. Figure 3 shows the probability of technique survival between the two disconnect systems. S142 Problems with dialysate bag [n (%)] 58 (33) 3 (2) Leakage 35 (20) 3 (2) Turbidity 4 (2) 0 (0) Protector 34 (19) 0 (0) Others 6 (3) 0 (0) Problems with Y-set [n (%)] 75 (43) 4 (2) Clamp 24 (14) 0 (0) Protector 21 (12) 0 (0) Leakage 3 (2) 0 (0) Cap 45 (26) 2 (1) Others 22 (13) 2 (1) Problems with drainage bag [n (%)] 36 (21) 10 (6) Leakage 23 (13) 10 (6) Clamp 5 (3) 0 (0) Others 11 (6) 0 (0) Total a 106 (60%) 15 (8%) a Because of overlap in the problems, the total is not equivalent to the sum of the individual events. DISCUSSION The Carex system from B. Braun Carex was introduced as a cheaper alternative to the existing Ultra system from Baxter Healthcare. Because of limitations in our health system, we were not able to conduct a randomized, parallel, controlled trial on new patients. Our approach was to conduct the study on prevalent patients. The resulting design has inherent problems. Natural attrition of patients may occur during the wait for conversion. Also, the product under study requires a conversion procedure and an adjustment to the new system that are not required in the control group. Not surprisingly, technical difficulties in using the system were reported more frequently with Carex. Product defects were also detected more frequently in the Figure 3 Probability of technique survival by disconnect system. Log-rank test χ 2 (1) = 19.14; p value < 0.0001.

PDI DECEMBER 2003 VOL. 23, SUPPL 2 PROCEEDINGS OF THE FIRST ASIAN CHAPTER MEETING ISPD Carex system. The trend toward fewer technical difficulties in the later periods of the study is probably attributable to adaptation by the patients to the system and to ongoing improvements in product design. The technical difficulty information should be interpreted with care, because we relied on self-reporting by patients. Independent verification of the events could not be obtained. With regard to the risk of peritonitis, we noted significant treatment-center interaction: that is, the risk of peritonitis on the Carex system as compared with the same risk on the Ultra system varied significantly between the centers. The reasons for this phenomenon are unknown. To provide an overall estimate of the risk of peritonitis with the Carex system, we used the fixed Poisson model. Using that model, equivalence could not be demonstrated. Although the study was slightly underpowered to detect equivalence, the results suggest that Carex was the inferior product, with a two-fold increase in risk of peritonitis. However, caution needs to be exercised in interpreting the combined results. Controversy exists over how center estimates may be combined, and different methods have different justifications and may yield inconsistent results (7,8). The observed difference in the rate of peritonitis was most likely attributable to the CAPD system. The randomization would have cancelled host risk factors out (9). However, data on Staphylococcus aureus nasal carriage and exit-site infection were not captured. Another possible confounding factor is the impact on the risk of peritonitis of a patient s learning curve with the new system. We could not find such an association reported in the literature. Overreporting or underreporting bias is unlikely. Reported peritonitis episodes were subject to independent and blinded verification. The proportion of peritonitis episodes not verified by that process was slightly higher in the control group, suggesting that reporting of peritonitis was not biased against the new system. CONCLUSIONS The therapeutic equivalence of the Carex CAPD disconnect system with the Ultra system could not be demonstrated. The risk of peritonitis with Carex was variable between centers, with significant center effect. ACKNOWLEDGMENT The authors thank B. Braun Malaysia for support of this research. We acknowledge the help of Day-Guat Lee of Hospital Kuala Lumpur. We thank the Clinical Research Centre, Hospital Kuala Lumpur, for overseeing and auditing the trial. We thank the Director General of Health Malaysia for permission to publish this paper. REFERENCES 1. Bazzato G, Landini S, Coli U, Lucatello S, Fracasso A, Moracchiello M. A new technique of continuous ambulatory peritoneal dialysis (CAPD): double-bag system for freedom to the patient and significant reduction of peritonitis. Clin Nephrol 1980; 13:251 4. 2. Maiorca R, Cantaluppi A, Cancarini GC, Scalamogna A, Broccoli R, Graziani G, et al. Prospective controlled trial of a Y-connector and disinfectant to prevent peritonitis in continuous ambulatory peritoneal dialysis. Lancet 1983; 2:642 4. 3. Canadian CAPD Clinical Trials Group. Peritonitis in continuous ambulatory peritoneal dialysis (CAPD): a multi-centre randomized clinical trial comparing the Y connector disinfectant system to standard systems. Perit Dial Int 1989; 9:159 63. 4. Burkart JM, Hylander B, Durnell Figel T, Roberts D. Comparison of peritonitis rates during long-term use of standard spike versus Ultraset in continuous ambulatory peritoneal dialysis (CAPD). Perit Dial Int 1990; 10:41 3. 5. Ryan P. Random allocation of treatments in blocks. In: Newton HJ, ed. Stata Technical Bulletin Reprints. College Station, TX: Stata Corporation; 1998: 7. 6. Jennison C, Turnbull BW. Repeated confidence intervals for group sequential clinical trials. Control Clin Trials 1984; 5:33 45. 7. Agresti A, Hartzel J. Strategies for comparing treatments on a binary response with multi-centre data. Stat Med 2000; 19:1115 39. 8. DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin Trials 1986; 7:177 88. 9. Burkhart J. Pathophysiology and prevention of peritonitis in continuous peritoneal dialysis. UpToDate 2000; 8.3.20. [Online at www.uptodate.com; subscription required] S143