What are the appropriate methods for analyzing patient-reported outcomes in randomized trials when data are missing?

Subjective health measurements using Patient Reported Outcomes (PRO) are increasingly used in randomized trials, particularly for patient groups comparisons. Two main types of analytical strategies can be used for such data: Classical Test Theory (CTT) and Item Response Theory models (IRT). These two strategies display very similar characteristics when data are complete, but in the common case when data are missing, whether IRT or CTT would be the most appropriate remains unknown and was investigated using simulations. We simulated PRO data such as quality of life data. Missing responses to items were simulated as being completely random, depending on an observable covariate or on an unobserved latent trait. The considered CTT-based methods allowed comparing scores using complete-case analysis, personal mean imputations or multiple-imputations based on a two-way procedure. The IRT-based method was the Wald test on a Rasch model including a group covariate. The IRT-based method and the multiple-imputations-based method for CTT displayed the highest observed power and were the only unbiased method whatever the kind of missing data. Online software and Stata® modules compatibles with the innate mi impute suite are provided for performing such analyses. Traditional procedures (listwise deletion and personal mean imputations) should be avoided, due to inevitable problems of biases and lack of power.

Mupirocin/chlorexidine to prevent methicillin-resistant Staphylococcus aureus infections: post hoc analysis of a placebo-controlled, randomized trial using mupirocin/chlorhexidine and polymyxin/tobramycin for the prevention of acquired infections in intubated patients

PURPOSE

The reduction in acquired infections (AI) due to methicillin-resistant Staphylococcus aureus (MRSA) with the mupirocin/chlorhexidine (M/C) decontamination regimen has not been well studied in intubated patients. We performed post hoc analysis of a prior trial to assess the impact of M/C on MRSA AI and colonization.

METHODS

We conducted a multicenter, placebo-controlled, randomized, double-blind study with the primary aim to reduce all-cause AI. The two regimens used [topical polymyxin and tobramycin (P/T), nasal mupirocin with chlorhexidine body wash (M/C), or corresponding placebos for each regimen] were administered according to a 2 × 2 factorial design. Participants were intubated patients in the intensive care units of three French university hospitals. The patients enrolled in the study (n = 515) received either active P/T (n = 130), active M/C (n = 130), both active regimens (n = 129), or placebos only (n = 126) for the period of intubation and an additional 24 h. The incidence and incidence rates (per 1,000 study days) of MRSA AI were assessed. Due to the absence of a statistically significant interaction between the two regimens, analysis was performed at the margins by comparing all patient receiving M/C (n = 259) to all patients not receiving M/C (n = 256), and all patients receiving P/T (n = 259) to all patients not receiving P/T (n = 256).

RESULTS

Incidence [odds ratio (OR) 0.39, 95 % confidence interval (CI) (0.16-0.96), P = 0.04] and incidence rates [incidence rate ratio (IRR) 0.41, 95 % CI 0.17-0.97, P = 0.05] of MRSA AI were significantly lower with the use of M/C. We also observed an increase in the incidence (OR 2.50, 95 % CI 1.01-6.15, P = 0.05) and the incidence rate (IRR 2.90, 95 % CI 1.20-8.03, P = 0.03) of MRSA AI with the use of P/T.

CONCLUSION

Among our study cohort of intubated patients, the use of M/C significantly reduced MRSA AI.

Power analysis on the time effect for the longitudinal Rasch model

Statistics literature in the social, behavioral, and biomedical sciences typically stress the importance of power analysis. Patient Reported Outcomes (PRO) such as quality of life and other perceived health measures (pain, fatigue, stress,...) are increasingly used as important health outcomes in clinical trials or in epidemiological studies. They cannot be directly observed nor measured as other clinical or biological data and they are often collected through questionnaires with binary or polytomous items. The Rasch model is the well known model in the item response theory (IRT) for binary data. The article proposes an approach to evaluate the statistical power of the time effect for the longitudinal Rasch model with two time points. The performance of this method is compared to the one obtained by simulation study. Finally, the proposed approach is illustrated on one subscale of the SF-36 questionnaire.

Ammonia plasma concentration and prolonged infusion of remifentanil in patients with acute kidney injury

BACKGROUND

Glycine is an excipient of remifentanil and may induce side effects. To investigate glycine and ammonia concentration with the use of remifentanil in Intensive Care Unit patients with acute kidney injury (AKI) defined by a decrease in creatinine clearance above 50%.

METHODS

Prospective open-label cohort study in three surgical Intensive Care Units. Thirty-three patients with AKI and requiring sedation for at least 72 hours. Sedation with remifentanil and midazolam or propofol was adapted every six hours according to ATICE. Glycine and ammonia plasma concentrations were measured at H0 (start of infusion) and every 12 hours during a continuous intravenous 72 hours remifentanil infusion, and 24 hours after the end of the infusion. Clinical and biological glycine or ammonia toxicity were evaluated.

RESULTS

Fifteen patients required continuous veno-venous hemodiafiltration (CVVHDF). Glycine and ammonia plasma concentrations exceeded the normal value respectively for 11 (33%) and 15 (45%) patients before remifentanil infusion (H0). Accumulation of glycine or ammonia was observed neither for patients with or without CVVHDF. For patients without CVVHDF, the plasma ammonia concentration at the end of remifentanil infusion was significantly correlated with the creatinine clearance at H72 (P=0.03) and with the mean rate of remifentanil infusion (P=0.002). No side effect was reported.

CONCLUSION

Remifentanil was not associated with an accumulation of glycine or ammonia in patients with AKI. Plasma ammonia concentration was correlated with the mean rate of remifentanil and creatinine clearance. A 72-hours remifentanil infusion appeared safe for sedation of patients with AKI.

Clinical trial: Radiofrequency energy delivery in proton pump inhibitor-dependent gastro-oesophageal reflux disease patients

BACKGROUND

Radiofrequency (RF) energy delivery is an endoscopic procedure developed for the treatment of gastro-oesophageal reflux disease.

AIM

To compare RF and a proton pump inhibitor strategy (PPI) in PPI-dependent patients by carrying out a prospective, randomized trial.

METHODS

Patients with PPI-dependent typical reflux symptoms were randomly allocated to either RF or PPI regimen alone. The primary endpoint, evaluated at 6-month, was defined as the possibility for the patient to stop or to decrease PPI use to <50% of the effective dose required at baseline.

RESULTS

In the RF group, 18/20 patients stopped (n = 3) or decreased (n = 15) PPI use as compared to eight of 16 in the PPI group (P = 0.01). None of the control patients could stop PPI. Health-related quality of life scores were not different between groups. No significant change in oesophageal acid exposure (OAE) was noted between baseline and 6-months after RF. No severe complication was reported.

CONCLUSIONS

Radiofrequency energy delivery is a safe and effective therapeutic option, allowing reduction in or discontinuation of PPI therapy in patients with PPI-dependent symptoms, without loss of quality of life. However, in a majority of patients, PPI therapy cannot be completely stopped. The efficacy of RF does not seem to be related to a decrease in OAE.

A prospective longitudinal study of BK virus infection in 104 renal transplant recipients

BK virus (BKV) infection during the first year after renal transplantation was studied prospectively in 104 unselected consecutive patients. Viral DNA in urine (DNAuria) and plasma (DNAemia) samples was detected and quantified by real-time PCR. The noncoding control region (NCCR) of BKV isolates was sequenced. DNAuria and DNAemia occurred in 57% and 29% of patients, respectively. Three groups were defined, uninfected patients (group 1, n=45), patients with DNAuria (group 2, n=29) and patients with positive DNAemia (group 3, n=30). Active infection started within the first 3 months in 80% of patients. Cold ischemia duration over 24 h and the administration of tacrolimus were identified as significant risks factors for DNAuria, whereas it remains more frequently negative in patients receiving cyclosporine A. The risk for positive DNAemia was higher in patients with DNAuria (notably for viral load (VL)>4 log/mL) or treated with tacrolimus. No relationship was found with genetic variability in the NCCR sequence. Our data highlight the high frequency of active BKV infection after renal transplantation. Although high VL was detected in some patients, none developed a BKV nephropathy. A prospective follow-up of the whole population during the first year post renal transplantation is thus not useful to predict BKV disease.

Non-invasive assessment of hepatic iron stores by MRI

BACKGROUND

MRI has been proposed for non-invasive detection and quantification of liver iron content, but has not been validated as a reproducible and sensitive method, especially in patients with mild iron overload. We aimed to assess the accuracy of a simple, rapid, and easy to implement MRI procedure to detect and quantify hepatic iron stores.

METHODS

Of 191 patients recruited, 17 were excluded and 174 studied, 139 in a study group and 35 in a validation group. All patients underwent both percutaneous liver biopsy with biochemical assessment of hepatic iron concentration (B-HIC) and MRI of the liver with various gradient-recalled-echo (GRE) sequences obtained with a 1.5 T magnet. Correlation between liver to muscle (L/M) signal intensity ratio and liver iron concentration was calculated. An algorithm to calculate magnetic resonance hepatic iron concentration (MR-HIC) was developed with data from the study group and then applied to the validation group.

FINDINGS

A highly T2-weighted GRE sequence was most sensitive, with 89% sensitivity and 80% specificity in the validation group, with an L/M ratio below 0.88. This threshold allowed us to detect all clinically relevant liver iron overload greater than 60 micromol/g (normal value <36 micromol/g). With other sequences, an L/M ratio less than 1 was highly specific (>87%) for raised hepatic iron concentration. With respect to B-HIC range analysed (3-375 micromol/g), mean difference and 95% CI between B-HIC and MR-HIC were quite similar for study and validation groups (0.8 micromol/g [-6.3 to 7.9] and -2.1 micromol/g [-12.9 to 8.9], respectively).

INTERPRETATION

MRI is a rapid, non-invasive, and cost effective technique that could limit use of liver biopsy to assess liver iron content. Our MR-HIC algorithm is designed to be used on various magnetic resonance machines.

Comparison of four sequential methods allowing for early stopping of comparative clinical trials

Phase III trials aim to assess whether a new treatment has superior efficacy than a standard treatment. Sequential methods, such as the sequential probability ratio test (SPRT), the triangular test (TT) and so-called one-parameter boundaries (OPB), now allow early stopping of such trials, both in the case of efficacy (alternative hypothesis; H(1)) and in the case of lack of efficacy (null hypothesis; H(0)). We compared the statistical properties of the SPRT and the TT, and of OPB with Pocock (OPB(Delta=0.5)) and O'Brien and Fleming (OPB(Delta=0)) type boundaries, in the setting of one-sided comparative trials with normal response. We studied the type I error (alpha), power (1-beta), average sample number (ASN) and 90th percentile (P90) of the number of patients required to reach a conclusion using simulations. The four tests were also compared with the corresponding single-stage design (SSD). All sequential tests display alpha and 1-beta close to nominal values and, as compared with SSD, allow important decreases in ASN: for example, -48%, -42%, -40% and -31% under H(0) and H(1) for SPRT, TT, OPB(Delta=0.5) and OPB(Delta=0) respectively. For situations between H(0) and H(1), ASNs of all sequential tests were still smaller than the sample size required by SSD, with the TT displaying the largest decrease (-25%). The P90s of the TT and OPB(Delta=0) under H(0) and H(1) were smaller than the P90s of the SPRT and OPB(Delta=0.5), which were similar to the sample size required by SSD. If all sequential tests display approximately similar features, the TT is the most appealing regarding decreases in sample size, especially for situations between H(0) and H(1).

Impaired pressor sensitivity to noradrenaline in septic shock patients with and without impaired adrenal function reserve

AIMS

To investigate the relationship between adrenal gland function and pressor response to noradrenaline in septic shock.

METHODS

Basal cortisol level, noradrenaline--mean arterial pressure dose-response curve and cortisol response to intravenous corticotrophin bolus were obtained in nine patients fulfilling usual criteria for septic shock and in six normal volunteers. In patients with septic shock, dose-response curve to noradrenaline was determined a second time 60 min after a 50 mg intravenous hydrocortisone bolus.

RESULTS

As compared with controls, patients with septic shock had increased basal cortisol levels (mean+/-s.d.: 1564+/-818 vs 378+/-104 nmol l(-1) , P=0.002, 95% confidence interval for difference in means: [452, 1920]) and a blunted cortisol response to corticotrophin (403+/-461 vs 1132+/-195 nmol l(-1), P=0.008, [-1163, -2951). Five patients had impaired adrenal function reserve. As compared with controls, septic patients displayed a moderate and non significant decrease in pressor sensitivity to noradrenaline (P=0.112). As compared with patients with adequate adrenal response, patients with impaired adrenal function reserve showed a significant decrease in pressor sensitivity to noradrenaline (P=0.038). In septic patients, hydrocortisone improved pressor response to noradrenaline (P=0.032). This effect was more marked in patients with impaired adrenal function reserve so that, as compared with patients with adequate response, the difference was no longer significant (P=0.123).

CONCLUSIONS

In septic shock, impaired adrenal function reserve may partly be accounted for by the depressed pressor sensitivity to noradrenaline. The latter may be substantially improved by physiological doses of hydrocortisone.

Effect of subcutaneous tunneling on internal jugular catheter-related sepsis in critically ill patients: a prospective randomized multicenter study

OBJECTIVE

To evaluate the effect of catheter tunneling on internal jugular catheter-related sepsis in critically ill patients.

DESIGN

A prospective randomized controlled study involving 3 intensive care units (ICUs), stratified by number of catheter lumina (1 or 2) and center.

SETTING

The 10-bed medical-surgical and 10-bed surgical ICUs at Saint Joseph Hospital and 8-bed surgical ICU at Clinique de la Défense, Paris, France.

PATIENTS

Every patient older than 18 years admitted to the ICUs between March 1, 1993, and July 17,1996, who required a jugular venous catheter for more than 48 hours.

INTERVENTION

Random allocation to tunneled or nontunneled catheters.

MEASUREMENTS

Times to occurrence of systemic catheter-related sepsis, catheter-related septicemia, or a quantitative catheter-tip culture with a cutoff of 103 colony-forming units per milliliter.

RESULTS

A total of 241 patients were randomized. Ten patients in whom jugular puncture was not achieved were subsequently excluded. The proportion of patients receiving mechanical ventilation (87%) and mean+/-SD age (65+/-4 years), Simplified Acute Physiologic Score (13.3+/-4.9), Organ System Failure score (1.5+/-1.0), and duration of catheterization (8.7+/-5.0 days) were similar in both groups. Taking into account the first 231 catheters (114 nontunneled [control], 117 tunneled), we found that tunnelization decreased catheter-related sepsis (odds ratio [OR], 0.33; 95% confidence interval [CI], 0.13-0.83; P=.02), catheter-related septicemia (OR, 0.23; 95% CI, 0.07-0.81; P=.02), and, though not statistically significant, positive quantitative tip-culture rate (OR, 0.62; 95% CI, 0.35-1.10; P=.10). These results were slightly modified after adjustment on parameters either imbalanced between both groups (duration of catheter placement and cancer at admission) or prognostic (insertion by a resident, use of antibiotics at catheter insertion, cancer, and sex).

CONCLUSION

The incidence of internal jugular catheter-related infections in critically ill patients can be reduced by using subcutaneous tunnelization.