Chest ACCP Member Benefits
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

Guest Access | Sign In via User Name/Password
This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF) Free
Right arrow Submit a response
Right arrow Alert me when this article is cited
Right arrow Alert me when eLetters are posted
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Article Archive
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via ISI Web of Science (19)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Mahon, J. L.
Right arrow Articles by Donner, A.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Mahon, J. L.
Right arrow Articles by Donner, A.
(Chest. 1999;115:38-48.)
© 1999 American College of Chest Physicians

Theophylline for Irreversible Chronic Airflow Limitation*

A Randomized Study Comparing n of 1 Trials to Standard Practice

Jeffrey L. Mahon, MD, MSc; Andreas Laupacis, MD, MSc; Richard V. Hodder, MD, MSc, FCCP; Douglas A. McKim, MD; Nigel A. M. Paterson, MD, FCCP; Thomas E. Wood, MD, MSc, FCCP and Allan Donner, PhD

* From the Department of Medicine (Drs. Mahon, Paterson, and Wood) and the Department of Epidemiology and Biostatistics (Dr. Donner), University of Western Ontario; and Department of Medicine (Drs. Laupacis, Hodder, and McKim), University of Ottawa. Canada.


   Abstract
TOP
 Abstract
Introduction
 Materials and Methods
 Interventions
 Outcome Measures and Sample...
 Analysis
 Assignment and Masking
 Results
 Discussion
 References
 
Study objective: To compare quality of life and exercise capacity (primary aim), and drug usage (secondary aim), between groups of patients with irreversible chronic airflow limitation (CAL) who were undergoing theophylline (Theo-Dur; Key Pharmaceuticals; Kenilworth, NJ) therapy guided by n of 1 trials or standard practice.

Design: Randomized study of n of 1 trials vs standard practice.

Setting: Outpatient departments in two tertiary care centers.

Patients: Sixty-eight patients with irreversible CAL who were symptomatic despite the use of inhaled bronchodilators, and who were unsure whether theophylline was helping them following open treatment, were randomized into n of 1 trials (N = 34) or standard practice.

Interventions: The n of 1 trials (single-patient, randomized, double-blind, multiple crossover comparisons of the effect on dyspnea of theophylline vs a placebo) followed published guidelines. Standard practice patients stopped taking theophylline but resumed it if their dyspnea worsened. If their dyspnea then improved, theophylline was continued. In both groups, a decision about continuing or stopping the use of theophylline was made within 3 months of randomization.

Measurements and results: The primary outcomes (the chronic respiratory disease questionnaire [CRQ] and 6-min walk) were measured at baseline, 6 months, and 12 months by personnel blinded to treatment group allocation. No between-group differences (n of 1 minus standard practice) were seen in within-group changes over time (1 year minus baseline) in the CRQ Physical Function score (point estimate on the difference, –2.8; 95% confidence limits [CLs], –8.2, 2.5), CRQ Emotional Function score (point estimate on the difference, 0.5; 95% CLs, –4.7, 5.7), or 6-min walk (point estimate on the difference, 8 m; 95% CLs, –26, 44 m). No differences between groups were seen in the secondary outcome of the proportion of patients taking theophylline at 6 and 12 months. In 7 of 34 n of 1 trial patients (21%), dyspnea improved during theophylline treatment compared with placebo treatment.

Conclusions: Using n of 1 trials to guide theophylline therapy in patients with irreversible CAL did not improve their quality of life or exercise capacity, or reduce drug usage, over 1 year compared to standard practice. Under the objective conditions of an n of 1 trial, 21% of patients with CAL responded to theophylline. There remains a rationale for considering theophylline in patients with irreversible CAL who remain symptomatic despite the use of inhaled bronchodilators, but the use of n of 1 trials to guide this decision did not yield clinically important advantages over standard practice.

Key Words: irreversible chronic airflow limitations • n of 1 trials • randomized trial • theophylline


   Introduction
TOP
 Abstract
 Introduction
Materials and Methods
 Interventions
 Outcome Measures and Sample...
 Analysis
 Assignment and Masking
 Results
 Discussion
 References
 
The role of theophylline for patients with irreversible chronic airflow limitation (CAL) remains unclear despite decades of experience.1 ,2 Theophylline has several potentially useful effects in these patients.3 However, randomized placebo-controlled trials have not consistently shown that theophylline improves their quality of life or exercise capacity.4 ,5 ,6 ,7 ,8 ,9 ,10 ,11 ,12 ,13 ,14 ,15 ,16 ,17 Small samples, the use of incompletely characterized quality-of-life indexes, and the failure to assess exercise capacity may partly explain the discordance. Thus, theophylline improved validated quality-of-life measures in three trials,7 ,9 ,14 improved dyspnea according to a visual analog scale in the largest trial,10 and improved exercise capacity in three trials.9 ,14 ,15 These positive results suggest that theophylline is effective in at least some patients with CAL and support recommendations to consider its use for those patients who remain symptomatic despite the use of inhaled bronchodilators.3 ,18 In this case, the usual clinical approach is a before-after therapeutic trial, in which theophylline is prescribed openly and its effectiveness is judged by the patient's subsequent responses using subjective (eg, dyspnea) and objective (eg, pulmonary function tests and/or exercise capacity) parameters.18

Open therapeutic trials in single patients carry inherent biases. These include the placebo effect, the tendency for physicians and patients to want and to expect the therapy to work, and regression to the mean.19 ,20 These biases probably favor false conclusions that a treatment is effective and can, therefore, be expected to lead to the unnecessary use of a therapy in some patients.19 ,20 ,21 As well, many chronic diseases including irreversible CAL show fluctuations in severity that cannot be explained clearly by extraneous events. This variability in "natural history" may result in false conclusions that a treatment is not effective in some patients if it is started simultaneously with a period of decline. The n of 1 trial (randomized, double-blind, multiple crossover comparisons of active and placebo treatments in a single patient) is a potential way to limit such biases.19 ,20 ,21

We hypothesized that the use of n of 1 trials to guide theophylline therapy in a group of patients with CAL would lead to a better outcome than that a group in which theophylline use was guided by open therapeutic trials.21 This improvement would result from more reliable identification of those patients who do, and do not, benefit from the drug. Randomization, blinding, and multiple crossovers would limit false conclusions that theophylline is effective by controlling for the placebo effect, patient and physician expectations, and regression to the mean; multiple crossovers would also limit false conclusions that theophylline is not effective by controlling for temporal variation in the underlying course of the patients' CAL. To the extent that the biases of open trials appear weighted toward false conclusions of efficacy, we also predicted that n of 1 trials would result in less overall theophylline use than standard practice.

In a previous randomized study of theophylline use for irreversible CAL guided by n of 1 trials vs standard practice, we found that n of 1 trials significantly reduced theophylline use over 6 months.21 However, improved quality of life and exercise capacity among n of 1 trial patients was not seen, possibly because of the small sample size.21 We report a larger study with follow-up to 1 year that tested two a priori hypotheses. Our primary hypothesis was that patients with irreversible CAL who were given theophylline guided by n of 1 trials would have better quality of life and exercise capacity than patients treated according to standard practice. Our secondary hypothesis was that n of 1 trials would result in less theophylline use than standard practice. We also report the results of a post hoc analysis in which patients from this and our previous study were combined.


   Materials and Methods
TOP
 Abstract
 Introduction
 Materials and Methods
Interventions
 Outcome Measures and Sample...
 Analysis
 Assignment and Masking
 Results
 Discussion
 References
 
Protocol
Our rationale for studying theophylline for irreversible CAL was twofold: (1) prescription of theophylline for CAL meets the prerequisites for n of 1 trials 19 ,20 (CAL is chronic and relatively stable, the action of theophylline starts and ends within several days, and the efficacy of theophylline in a specific patient with CAL is often in doubt); and (2) successful n of 1 trials of theophylline for CAL have been reported.22 Approval by institutional review boards was obtained, and all patients provided informed consent. The diagnosis of irreversible CAL required: (1) an FEV1 of < 70% predicted and an FEV1/FVC ratio of < 70% predicted on two occasions within 1 month; and (2) an increase in FEV1 of <= 15% and <= 200 mL after the use of inhaled salbutamol. Patients were recruited from four outpatient clinics in two tertiary care Canadian centers. Two strata were defined before randomization. "Prior-theophylline-use" patients were taking theophylline at the time of recruitment and were using inhaled bronchodilators (ipratropium bromide, 40 µg qid, salbutamol, 200 µg bid, or another inhaled ß2-agonist in adequate dosage). "No-prior-theophylline-use" patients were also using inhaled bronchodilators but were not taking theophylline at the time of recruitment. Before being entered into the study, these patients were prescribed open-label theophylline (Theo-Dur; Key Pharmaceuticals; Kenilworth, NJ), to be used for at least 2 weeks, and the dose was adjusted as required to obtain theophylline blood levels between 30 and 110 µmol/L without side effects.

Patients were excluded before randomization if (1) they could not tolerate theophylline at any dose, or (2) they answered "yes" to the question, "Are you certain that theophylline is helping you?" while having theophylline blood levels between 30 and 110 µmol/L. These exclusions arose from the prediction that n of 1 trials are of little use in patients who demonstrate obvious beneficial effects or adverse effects during an open run-in with the treatment.23 In both cases, n of 1 trials are likely to simply confirm these responses and are, therefore, a waste of time and effort.23


   Interventions
TOP
 Abstract
 Introduction
 Materials and Methods
 Interventions
Outcome Measures and Sample...
 Analysis
 Assignment and Masking
 Results
 Discussion
 References
 
The n of 1 trials followed published guidelines (Fig 1 ).23 Patients identified daily activities during which they could become dyspneic. These activities were incorporated into a diary, in which the severity of dyspnea was rated on a 7-point Likert scale. For example, if dyspnea while climbing stairs was identified, the question read "How short of breath were you today when you were climbing stairs?," with response options of 1 (extremely short of breath), 2 (very short of breath), 3 (quite a bit short of breath), 4 (moderately short of breath), 5 (mildly short of breath), 6 (a little short of breath), and 7 (not at all short of breath). Sustained-release theophylline and an identical placebo were dispensed in pairs of 10-day periods (ie, 10 days for theophylline and then 10 days for placebo). The order of theophylline and placebo within pairs was randomly determined by a computer-generated scheme held in pharmacies at each site; the physicians supervising the n of 1 trials were also unaware of which drug the patient was taking. For patients receiving theophylline prior to randomization (all no-prior-theophylline-use patients and all prior-theophylline-use patients taking theophylline at the time of recruitment), the starting dose of theophylline was the same as that used before randomization. For prior-theophylline-use patients who were receiving a preparation containing oxtriphylline or aminophylline, the starting dose of theophylline was adjusted accordingly.



View larger version (22K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 1. Design of the n of 1 trials.

 
Diaries were completed on days 6 to 10, inclusive, for each treatment period to allow for drug washout from days 1 to 5. Patients could contact the study physician before completing a treatment period if their symptoms worsened. If the change could not be explained clinically by an event other than the withholding of theophylline (eg, a respiratory infection), patients switched to the other treatment.23 Patients were reviewed by the physician at the end of each treatment pair. A maximum of four pairs could occur. However, the patient or physician could stop the n of 1 trial earlier for any reason.

Immediately on stopping the n of 1 trial, the code was opened and the mean symptom score for each 10-day treatment period was determined. If the patient had received at least one treatment pair, the point estimate on the difference in mean symptom score within treatment pairs (theophylline minus placebo) was determined; if two or more treatment pairs were received, the 90% confidence limits (CLs) on this estimate were also determined.23 If diary scores were not available for at least one treatment pair, a recommendation to stop or continue theophylline was made according to clinical judgment and patient preference. For all other n of 1 trials, the recommendation about theophylline therapy was guided by a previously described algorithm that considered the point estimate on the mean difference in symptom score and, where available, its 90% CLs (Fig 2 ).21 We accepted 90% CLs, rather than the conventional 95% CLs, in deciding about a treatment effect within the n of 1 trials as a compromise between the inherent low power of n of 1 trials having four or less treatment pairs and the need to avoid a type I (false-positive) statistical error.20 ,21 Assumptions implicit in the algorithm were that the minimal important difference for n of 1 trials using a 7-point Likert scale was a mean change of 0.5/symptom24 and that within-patient serial correlation in mean responses could be ignored.25



View larger version (32K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 2. Interpretation of n of 1 trial results.

 
Patients treated according to standard practice stopped taking theophylline. They were asked to contact the study physician if their dyspnea worsened. Theophylline was resumed if this deterioration was deemed, on clinical grounds, not to be secondary to a new problem (eg, a respiratory infection). If their dyspnea improved on resuming theophylline, they were asked to remain on the drug. Standard practice patients were reviewed 3 months after randomization, when their status concerning theophylline therapy was recorded.

Patients in both groups returned to their primary care physicians once the initial 3-month phase of the study was completed. Changes to their treatment including changes regarding theophylline were left to the discretion of the primary care physician.


   Outcome Measures and Sample Size
TOP
 Abstract
 Introduction
 Materials and Methods
 Interventions
 Outcome Measures and Sample...
Analysis
 Assignment and Masking
 Results
 Discussion
 References
 
The chronic respiratory disease questionnaire (CRQ), 6-min walk, FEV1, and FVC were assessed at baseline, 6 months, and 12 months in both the n of 1 trial and standard practice groups. The CRQ is a valid, responsive, quality-of-life index specific for CAL 9 ,26 that measures four domains. These domains can be combined into larger domains of Physical Function (combining dyspnea and fatigue) and Emotional Function (combining emotional function and mastery).9 The 6-min walk has been used to measure exercise capacity in several diseases, including irreversible CAL, and was administered with standardized encouragement.9 ,27 ,28 The items and response scale (dyspnea during up to five daily activities and a 7-point Likert response scale) comprising the CRQ dyspnea domain, as measured at baseline, 6 months, and 12 months, were identical to those used in the personalized daily patient diaries that were used by patients randomized to n of 1 trials, with three exceptions. First, the n of 1 trial patients did not complete the other three CRQ domains when completing their diaries during the n of 1 trials. Second, the reference time for dyspnea as measured in the n of 1 trials was the day that the diary was being completed (eg, "How short of breath were you today when you were climbing stairs?") vs the preceding 2 weeks when completing the full CRQ at baseline, 6 months, and 12 months (eg, "Please indicate how much shortness of breath you have had in the last 2 weeks while climbing the stairs?"). Third, dypnea ratings during the n of 1 trials used patient self-report, whereas completion of the full CRQ was performed in the presence of a study nurse.

The recommendation within the first 3 months to continue or stop theophylline usage for all patients was recorded, as was the proportion of patients taking theophylline at 6 and 12 months. Patient and physician confidence in the treatment decisions were independently assessed by the question, "How confident are you that theophylline should be stopped?" or "... continued?," with responses ranging from 1 (totally comfortable) to 7 (extremely uncomfortable). Cointerventions (inhaled agents [beclamethasone diproprionate, budesonide, flunisolide, ß2-agonists, and ipratropium bromide], prednisone, and home oxygen) in the week before the 6- and 12-month assessments were also recorded.

A sample of 80 patients was chosen based on a two-tailed {alpha}-error of 5%, 80% power, a 20% loss to follow-up, and the use of the CRQ Physical and Emotional function scores and the 6-min walk as primary outcomes. Minimal important differences for within-group changes over 1 year were assumed to be 4.5 for the Physical Function score,29 5.5 for the Emotional Function score,29 and 30 m for the 6-min walk.9 ,27 Estimates for within-group variance over time for these measures were derived from our previous trial.21 An a priori sample size estimate for the differences in drug usage between the two groups was not done because it was a secondary hypothesis.


   Analysis
TOP
 Abstract
 Introduction
 Materials and Methods
 Interventions
 Outcome Measures and Sample...
 Analysis
Assignment and Masking
 Results
 Discussion
 References
 
Two-way analyses of variance, accounting for the randomization strata and n of 1 trial or standard practice therapy, were performed. The effects of interest were the changes from baseline at 6 and 12 months in the CRQ Physical and Emotional Function scores and in the 6-min walk. The primary analysis compared all n of 1 trial patients to all standard practice patients. A prespecified secondary analysis compared n of 1 trial patients to standard practice patients within the prior-theophylline-use and no-prior-theophylline-use strata. The level of statistical significance was set at 5% (two-tailed) and was not adjusted for multiple comparisons. Similar analyses were performed for changes in FEV1 and FVC. Comparison of proportions between groups used the normal approximation to the binomial distribution, provided that the expected values within each cell of the contingency table exceeded 5.30 If at least one of the expected values was < 5, proportions were compared by Fisher's exact test. Ninety-five percent CLs for differences between groups were based on the normal approximation to the binomial distribution for proportions and Student's t distribution for means. Measures of patient and physician confidence in the treatment decision were assessed by the Mann-Whitney U test. All results shown have used an intent-to-treat rule and include all available data on patients withdrawn from the study to the time of withdrawal, unless otherwise indicated.


   Assignment and Masking
TOP
 Abstract
 Introduction
 Materials and Methods
 Interventions
 Outcome Measures and Sample...
 Analysis
 Assignment and Masking
Results
 Discussion
 References
 
Patients were individually randomized at each site to either n of 1 trials or standard practice. Computer-generated sequences were provided to the sites in sequentially numbered, opaque, sealed envelopes. Opening the envelope constituted randomization and was done by the study physicians immediately before patients began the n of 1 or standard practice trials. Personnel measuring the CRQ and 6-min walk were unaware of the treatment group allocation, and patients were instructed to maintain this blinding.


   Results
TOP
 Abstract
 Introduction
 Materials and Methods
 Interventions
 Outcome Measures and Sample...
 Analysis
 Assignment and Masking
 Results
Discussion
 References
 
Participant Flow and Follow-up
Seven patients placed openly on theophylline were not randomized because of side effects (five patients) or a perceived improvement in dyspnea (two patients) such that they did not want to stop taking theophylline (Fig 3 ). Sixty-eight patients were randomized to n of 1 trials or standard practice (34 patients each). The baseline characteristics of the groups appeared to be well matched (Table 1 ). Seven randomized patients (three n of 1 trial patients and four standard practice patients) were withdrawn: five patients died (two from end-stage respiratory failure and one each of complications following pulmonary volume-reduction surgery, recurrent small-cell carcinoma of the lung, and lymphoma); one patient required coronary artery angioplasty for unstable angina; and one patient was lost to follow-up after 1 month. Three other patients experienced major events that likely affected their quality of life and exercise capacity, but they completed the study (one patient in each group underwent elective total hip replacement, and one standard practice patient received chemotherapy for newly diagnosed lymphoma). Analyses excluding the 10 patients having major events did not yield different results from intent-to-treat analyses.



View larger version (47K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 3. Participant flow and follow-up.

 

View this table:
[in this window]
[in a new window]

  		Table 1. Baseline Characteristics

 
Comparsion of All n of 1 Trial Patients to All Standard Practice Patients (Primary Analysis)
Table 2 shows the CRQ scores, 6-min walk results, and spirometry values over time. No significant differences between the n of 1 trial patients and standard practice patients were seen in changes from baseline in these outcomes at 6 and 12 months. The 95% CLs for the between-group differences (n of 1 trial minus standard patients) in within-group changes over time (12 months minus baseline) in the CRQ Physical Function score, CRQ Emotional Function score, and 6-min walk were, respectively: –8.2, 2.5; –4.7, 5.7; and –26, 42 m).


View this table:
[in this window]
[in a new window]

  		Table 2. Mean (SD) Change from Baseline in CRQ, 6-Minute Walk, and Spirometry Values

 
Table 3 summarizes theophylline treatment over time. For n of 1 trial patients, theophylline status at 1 month was not recorded because the treatment decision was usually not made until at least the second month. No significant differences in theophylline use between the two groups were seen at any point over the year. In seven patients (five n of 1 trial patients and two standard practice patients), theophylline was subsequently resumed by nonstudy physicians despite conclusions during the first 3 months of the study that it was not relieving dyspnea. These reversals were associated with slightly lower, but nonsignificant, patient confidence in the original decision (median confidence level among patients who switched, 3 [range, 2 to 5]; median confidence level among patients who did not switch, 2 [range, 1 to 6]; p > 0.2). Treatment reversals in the other direction (ie, subsequently stopping theophylline usage following a decision during the first 3 months that it was effective) did not occur. No significant differences in home oxygen use or cointerventions were seen between the n of 1 trial and standard practice patients in the week preceding the 6- and 12-month visits (data not shown).


View this table:
[in this window]
[in a new window]

  		Table 3. Proportion of Patients Using Theophylline

 
The confidence of physicians and patients in treatment decisions about theophylline use did not differ, whether the decision was based on an n of 1 trial or on standard practice (median confidence level in decision for physicians, 2 [range, 1 to 4]; median confidence level in decision for patients, 2 [range, 1 to 6]). The confidence level of physicians was slightly higher following the n of 1 trial (median, 2; range, 1 to 4) than following standard practice (median, 3; range, 1 to 4). The confidence level of patients also was higher following the n of 1 trial (median, 2; range, 1 to 6) than following standard practice (median, 2.5; range, 1 to 6). However, these differences were not significant.

Comparison of n of 1 Trial and Standard Practice Patients According to Duration of Theophylline Use at Entry (Secondary Analysis)
Tables 2 and 3 also show the CRQ scores, 6-min walk results, and results of theophylline therapy over time among n of 1 trial patients and standard practice patients according to the prerandomization strata. An apparent interaction was present between previous theophylline use and theophylline treatment groups in the CRQ Emotional Function scores at 6 months (p = 0.034) and 12 months (p = 0.039). Comparisons between n of 1 trial patients and standard practice patients within the subgroups at 6 and 12 months were not significant, with the exception of an increase in the Emotional Function score among n of 1 trial patients ( + 5.6) compared with a decrease among standard practice patients (–2.2) at 6 months (p = 0.013, unpaired t test) within the no-prior-theophylline-use group.

The mean baseline theophylline level was higher in the prior-theophylline-use group than in the no-prior-theophylline-use groups (64 µmol/L vs 49 µmol/L, respectively; p = 0.002). A greater number of prior-theophylline-use patients than no-prior-theophylline-use patients were taking theophylline at 12 months (22/35 patients [63%] vs 9/26 patients [35%], respectively; p = 0.02). Among no-prior-theophylline-use patients, no differences were seen in the proportion of n of 1 trial patients and standard practice patients taking theophylline over time. Among prior-theophylline-use patients, more standard practice patients than n of 1 trial patients were taking theophylline at 3 months (26% more) and 6 months (21% more). However, these differences were not statistically significant (3 months, p = 0.18; 6 months, p = 0.19; Fisher's exact test).

Conduct and Results of n of 1 Trials
Four n of 1 trial patients stopped their n of 1 trials before or on completion of the first treatment period and refused to continue. Of these, three patients in the prior-theophylline-use group chose to remain on theophylline openly (two patients had received theophylline only and one patient had received placebo only), and one patient in the no-prior-theophylline-use group had received theophylline only and chose not to continue the drug. Of the remaining 30 patients, 4 provided diary scores for one pair of theophylline and placebo treatments from which a point estimate on the difference in dyspnea score could be determined. The other 26 patients completed at least two treatment pairs (8 patients, two pairs; 6 patients, three pairs; 12 patients, four pairs) from which CLs on the point estimate could also be determined. Early crossovers to alternate treatments within pairs occurred in six patients, and four patients were treated for a respiratory infection during the n of 1 trials but did not crossover or end the treatment period early.

Seven of 34 patients (21%) undergoing n of 1 trials demonstrated a potentially useful improvement in dyspnea while on theophylline (Table 4 ). In each case, the point estimate on the improvement in mean daily dyspnea score, as measured by the 7-point Likert scale during theophylline therapy, exceeded zero (range, 0.3 to 3.1), and the 90% CLs on the estimates were either removed from zero (five patients) or marginally overlapped zero (two patients). Seven of 18 prior-theophylline-use patients (39%) and 1 of 13 no-prior-theophylline-use patients (8%) (p = 0.10; Fisher's exact test) chose either to resume theophylline following a negative n of 1 trial, or to abandon their n of 1 trial in favor of remaining on theophylline, even though they had not received at least one course of both theophylline and placebo treatments.


View this table:
[in this window]
[in a new window]

  		Table 4. Results of Selected n of 1 Trials

 
Post Hoc Analysis Combining This Study and Our Previous Study
Our previous study also randomized patients with irreversible CAL to guide theophylline therapy by n of 1 trials or standard practice.21 However, that study was limited to patients taking theophylline at the time of recruitment into the study, making them most comparable to prior-theophylline-use patients in the present study, and follow-up went only to 6 months.21 Therefore, we combined prior-theophylline-use patients in the present study with patients from our previous study (N = 61) and assessed their responses over 6 months. In this case, theophylline use was significantly higher at 6 months among standard practice patients (21/29 [72%]) than among n of 1 trial patients (13/32 [41%]): a 32% difference (standard practice minus n of 1 trial) and 95% CLs on the difference (9%, 56%). This difference was not associated with significant between-group changes from baseline to 6 months in the CRQ Physical Function score (point estimate in difference [n of 1 trial minus standard practice], 0.3; 95% CLs, –3.2, 3.9), CRQ Emotional Function score (point estimate in difference, 0.5; 95% CLs, –3.7, 4.6), or 6-min walk (point estimate in difference, –17 m; 95% CLs, –44, 11 m).


   Discussion
TOP
 Abstract
 Introduction
 Materials and Methods
 Interventions
 Outcome Measures and Sample...
 Analysis
 Assignment and Masking
 Results
 Discussion
References
 
When determining therapy in single patients, the objectivity of n of 1 trials makes them an appealing alternative to the standard approach of open, before-after trials. Despite this, n of 1 trials are rarely used, for two reasons. First, they make extra demands of patients and physicians. Second, randomized studies have not shown that n of 1 trials improve clinical outcomes over standard practice. The present study is the largest such randomized study to date and did not confirm our main hypothesis that prescribing theophylline to patients with irreversible CAL by n of 1 trials improved quality of life and exercise capacity compared to standard practice. Therefore, we cannot recommend routine use of n of 1 trials to guide decisions about theophylline treatment for this problem.

We doubt that biased assessment of quality of life and exercise capacity explains our negative result. Personnel ascertaining these measures were masked to treatment- group allocation. We also suspect that the isolated finding of a higher CRQ Emotional Function score at 6 months among n of 1 trial patients within the No Prior Theophylline Use subgroup reflected chance. Thus, other reasons for our negative finding, beyond the main explanation that n of 1 trials are not effective in this setting, should be considered.

First, the question of inadequate statistical power can be addressed by comparing the observed 95% CLs for the between-group differences (n of 1 trial minus standard patients) in within-group changes over time (12 months minus baseline) in the primary outcomes to their corresponding minimal important differences. Given the respective CLs and minimal important differences for the CRQ Physical Function score (–8.2, 2.5; 4.5),29 the CRQ Emotional Function score (–4.7, 5.7; 5.5), 29 and the 6-min walk (–26, 42 m; 30 m),9 ,27 it remains possible, but unlikely, that clinically important effects in favor of n of 1 trials were missed. A second possible reason for our negative result is that theophylline is not effective for irreversible CAL through 1 year. This question has not been answered in existing randomized, placebo-controlled trials because no studies have assessed theophylline's effect beyond 2 months.4 ,5 ,6 ,7 ,8 ,9 ,10 ,11 ,12 ,13 ,14 ,15 ,16 ,17 If the drug's impact on patient well-being is not maintained in the longer term, then our study could not have been expected to show a difference.

A third possible reason for the negative result was that the n of 1 trials were conducted suboptimally. Nearly one half (16/34) of the n of 1 trials were based on two or less pairs of active and placebo treatments, including four patients in whom only one course of either active or placebo treatment was received. Because the experimental unit in an n of 1 trial is a treatment period, the power to detect differences between two therapies depends directly on the number of treatment periods. Thus, n of 1 trials having very few treatment periods are unlikely to detect weak treatment effects. This tendency toward low power is accentuated by the recommendation, to which we adhered, of excluding patients who were sure that theophylline was helpful following an open trial.23 However, our rate of stopping one half of the n of 1 trials before completing at least three treatment pairs is consistent with other studies,20 ,31 and it suggests that the failure of patients to complete multiple treatment crossovers is an inherent limitation of n of 1 trials. Our decision to exclude patients who were sure that theophylline was helpful following open therapy should not have invalidated conclusions regarding the effects of n of 1 trials or standard practice on group outcomes because of differing baseline perceptions about the drug's value. This conclusion is based on our randomizing patients to n of 1 trials or standard practice only after confirming that they were uncertain about the effects of theophylline. However, this approach does limit the generalizability of our findings. Specifically, we did not assess the impact of initiating theophylline by either n of 1 trials or standard practice, and without an open run-in, in patients with CAL.

A final possible reason for the negative result may reflect our decision to return patients to their own physician following the first 3 months of the study. In so doing , we adopted an effectiveness ("real world") rather than efficacy ("ideal world") strategy to assess the impact of n of 1 trials. In seven patients, nonstudy physicians subsequently chose to resume theophylline despite conclusions during the n of 1 trials or standard practice approach that it was ineffective. We cannot determine the impact of these reversals on the comparisons of quality of life and exercise capacity between the n of 1 trial group and the standard practice group beyond noting that (1) there was not a large difference between the two groups in the number of patients who subsequently resumed theophylline (five and two patients, respectively), and (2) excluding these patients did not yield different results from the primary, intent-to-treat analysis. It is, therefore, less likely that these switches materially affected our overall findings. However, investigators comparing outcomes between patients randomized to n of 1 trials or standard practice in future studies should consider limiting subsequent treatment changes (by directly controlling patient therapy beyond the initial phase) or allowing such reversals to occur but maintaining study power to detect differences after excluding patients who switch (by randomizing a larger number) .

Our secondary hypothesis that n of 1 trials would lead to less theophylline use than standard practice was also not confirmed in the present study. It was based on the premise that, overall, the biases of single-patient open trials favor false conclusions that the treatment is effective, and our previous randomized study of 26 patients in which 47% (95% CLs, 14%, 80%) fewer n of 1 trial patients than standard practice patients were taking theophylline at 6 months without differences in CRQ scores and the 6-min walk.21 Patients in our previous study had been on theophylline at the time of recruitment for at least a year, making them more comparable to prior-theophylline-use patients in the present study. Both patient groups were probably "theophylline survivors," being less sensitive to the drug's side effects and having a stronger perception that it was relieving symptoms compared to no-prior-theophylline-use patients. This inference is supported by our observations of prior-theophylline-use patients versus no-prior-theophylline-use patients of higher theophylline levels at entry, a greater proportion of patients taking theophylline at 1 year, and a larger proportion of patients resuming theophylline following a negative n of 1 trial or prematurely abandoning the n of 1 trial in favor of continuing the drug.

Despite the greater tendency among prior-theophylline-use patients to tolerate and remain on theophylline, we found that and fewer n of 1 trial patients than standard practice patients within this stratum were taking theophylline at 3 months (26%) and 6 months (21%), without clinically important changes in CRQ scores and exercise capacity. While the reduction in theophylline use and the lack of changes in quality of life and exercise capacity did not reach statistical significance, given the risks of theophylline1 and the large number of patients currently on the drug, these findings would be of potential clinical and economic importance if confirmed in a larger sample. We explored this possibility by combining prior-theophylline-use patients in the present trial with patients from our previous trial. We found a statistically and clinically significant reduction in theophylline use (32%) at 6 months among n of 1 patients vs standard practice patients, without significant reductions in the CRQ Physical and Emotional Function scores and the 6-min walk. This finding supports the presence of an overall bias toward false conclusions of efficacy when stopping theophylline openly in patients with CAL who have been taking it for some time, and it provides justification to use n of 1 trials in this setting. However, this conclusion must be qualified because it is based on a post hoc analysis for which confirmation is needed.

Our study neither strengthens nor diminishes the case for considering a trial of theophylline in some patients with irreversible CAL. In particular, our overall negative result does not mean that theophylline should be abandoned because it is ineffective in all such patients. This is because 7 of 34 patients (21%) undergoing n of 1 trials demonstrated a potentially useful response to the drug on the basis of their daily dyspnea scores (Table 4 ). In each case, the improvement in dyspnea score on theophylline exceeded zero, and the 90% CLs on the estimates made it unlikely, though not impossible, that chance explained the difference. Therefore, given the objectivity of n of 1 trials, the improvement in dyspnea during theophylline therapy in these patients seems genuine. Furthermore, these results corroborate those of previous parallel-group, placebo-controlled trials in which theophylline improved the quality of life and/or exercise capacity.7 ,9 ,10 ,14 ,15 Thus, there remains a rationale for a trial of theophylline in selected patients with irreversible CAL who remain symptomatic despite inhaled bronchodilators.

In conclusion, guiding theophylline therapy for patients with irreversible CAL by n of 1 trials did not improve their quality of life or exercise capacity, or reduce overall drug usage through 1 year compared to standard practice. Although some patients with CAL who underwent n of 1 trials of theophylline noted an improvement in dyspnea, we did not show that routine use of n of 1 trials to establish the value of theophylline in this setting yielded important advantages compared to the standard practice of open, before-after trials. In a post hoc analysis requiring confirmation, using n of 1 trials in patients with CAL who were already taking theophylline reduced drug usage over 6 months compared to standard practice without changes in their quality of life or exercise capacity. Further randomized studies of n of 1 trials vs standard practice for this and other problems are needed before widespread adoption of n of 1 trials can be recommended.32


   Acknowledgements
 
ACKNOWLEDGMENT: We acknowledge the help of study nurses Pamela Wilton, Eda Athill, Katherine Brown, Alison McKenzie, Erika Timmins, and Kathy Walker; Cindy Cartwright-Cheval and Karen Weeks for secretarial help; and Drs. Brian Feagan, Ross Feldman, and Kelly Zarnke for reviewing the manuscript.


   Footnotes
 
This work was supported by the Ontario Ministry of Health, grant no. 04348F. Astra Pharma Incorporated (Mississauga, Ontario) provided the medication used in this study.

Correspondence to: Dr. Jeffrey Mahon, Rm 6OF11, University Hospital, PO Box 5339, 339 Windermere Rd, London, ON N6A 5A5, Canada; e-mail: Jl.Mahon@lhsc.on.ca

Abbreviations:CAL = irreversible chronic airflow limitations; CL = confidence limit; CRQ = chronic respiratory disease questionnaire

Received for publication December 22, 1997. Accepted for publication July 16, 1998.


   References
TOP
 Abstract
 Introduction
 Materials and Methods
 Interventions
 Outcome Measures and Sample...
 Analysis
 Assignment and Masking
 Results
 Discussion
 References
 

  1. Newhouse, MT (1990) Is theophylline obsolete? Chest 98,1-2[Free Full Text]
  2. Jenne, JW (1990) Theophylline is no more obsolete than "two puffs qid" of current beta-2 agonists. Chest 98,3-4[Free Full Text]
  3. . Canadian Thoracic Society Workshop Group. (1992) Guidelines for the assessment and management of chronic obstructive pulmonary disease. Can Med Assoc J 147,420-428[Abstract]
  4. Alexander, MR, Dull, WL, Kasik, JE (1980) Treatment of chronic obstructive pulmonary disease with orally administered theophylline. A double blind controlled study. JAMA 244,2286-2290[Abstract]
  5. Eaton, ML, Green, BA, Church, TR, et al (1980) Efficacy of theophylline in "irreversible" airflow obstruction. Ann Int Med 92,758-761
  6. Eaton, ML, MacDonald, FM, Church, TR, et al (1982) Effects of theophylline on breathlessness in patients with chronic airflow limitation. Chest 82,538-542[Abstract/Free Full Text]
  7. Mahler, DA, Matthay, RA, Snyder, PE, et al (1985) Sustained-release theophylline reduces dysnea in nonreversible obstructive airway disease. Am Rev Respir Dis 131,22-25[ISI][Medline]
  8. Dullinger, D, Kronenberg, R, Niewoehner, DE (1986) Efficacy of inhaled metaproterenol and orally-administered theophylline in patients with chronic airflow limitation. Chest 89,171-173[Abstract/Free Full Text]
  9. Guyatt, GH, Townsend, M, Pugsley, SO, et al (1987) Bronchodilators in chronic airflow limitation. Am Rev Respir Dis 135,1069-1074[ISI][Medline]
  10. Murciano, D, Auclair, M-H, Pariente, R, et al (1989) A randomized, controlled trial of theophylline in patients with severe chronic obstructive pulmonary disease. N Eng J Med 320,1521-1525[Abstract]
  11. Thomas, P, Pugsley, JA, Stewart, JH (1992) Theophylline and salbutamol improve pulmonary function in patients with irreversible chronic obstructive pulmonary disease. Chest 101,160-165[Abstract/Free Full Text]
  12. Nishimura, K, Koyama, H, Ikeda, A, et al (1993) Is oral theophylline effective in combination with both inhaled antichinergic agent and inhaled beta 2-agonist in the treatment of stable COPD? Chest 104,179-184[Abstract/Free Full Text]
  13. Mulloy, E, McNicholas, WT (1993) Theophylline improves gas exchange during rest, exercise, and sleep in severe chronic obstructive pulmonary disease. Am Rev Respir Dis 148,1030-1036[ISI][Medline]
  14. Kirsten, DK, Wegner, RE, Jorres, RA, et al (1993) Effects of theophylline withdrawl in severe chronic obstructive pulmonary disease. Chest 104,1101-1107[Abstract/Free Full Text]
  15. Newman, D, Tamir, J, Speedy, L, et al (1994) Physiological and neuropsychological effects of theophylline in chronic obstructive pulmonary disease. Isr J Med Sci 30,811-816[Medline]
  16. Iversen, ET, Eiken, PA, Nielsen, LM, et al (1994) Theophylline treatment of irreversible, chronic obstructive pulmonary disease. Ugeskr-Laeger 156,5696-5699
  17. Nishimura, K, Koyama, H, Ikeda, A, et al (1995) The additive effect of theophylline on a high-dose combination of inhaled salbutamol and ipratropium bromide in stable COPD. Chest 107,718-723[Abstract/Free Full Text]
  18. Ferguson, GT, Cherniack, RM (1993) Management of chronic obstructive pulmonary disease. N Engl J Med 328,1017-1022[Free Full Text]
  19. Guyatt, G, Sackett, D, Taylor, DW, et al (1986) Determining optimal therapy – randomized trials in individual patients. N Eng J Med 314,889-892[Abstract]
  20. Guyatt, GH, Keller, JL, Jaeschke, R, et al (1990) The n of 1 randomized controlled trial: clinical usefulness. Ann Intern Med 112,293-299
  21. Mahon, JL, Laupacis, A, Woods, T, et al (1996) Randomized trial of n of 1 trials versus standard practice. BMJ 312,1069-1074[Abstract/Free Full Text]
  22. Patel, A, Jaeschke, R, Guyatt, GH, et al (1991) Clinical usefulness of N-of-1 randomized controlled trials in patients with nonreversible chronic airflow limitation. Am Rev Respir Dis 144,962-964[ISI][Medline]
  23. Guyatt, G, Sackett, D, Adachi, J, et al (1988) A clinician's guide for conducting randomized trials in individual patients. Can Med Assoc J 139,497-503[Abstract]
  24. Jaeschke, R, Guyatt, G, Keller, J, et al (1991) Interpreting changes in quality-of-life score in N of 1 randomized trials. Controlled Clin Trials 12,226S-233S[CrossRef][Medline]
  25. Guyatt, GH, Heyting, A, Jaeschke, R, et al (1990) N of 1 randomized trials for investigating new drugs. Controlled Clin Trials 11,88-100[CrossRef][Medline]
  26. Guyatt, GH, Berman, LB, Townsend, M, et al (1987) A measure of quality of life for clinical trials in chronic lung disease. Thorax 42,773-778[Abstract]
  27. Guyatt, GH, Sullivan, MJ, Thompson, PJ, et al (1985) The six-minute walk: a new measure of exercise capacity in patients with chronic heart failure. Can Med Assoc J 132,919-923[Abstract]
  28. Guyatt, GH, Pugsley, SO, Sullivan, M, et al (1984) Effect of encouragement on walking test performance. Thorax 39,818-822[Abstract]
  29. Jaeschke, R, Singer, J, Guyatt, GH (1989) Measurement of health status. Ascertaining the minimal clinically important difference. Controlled Clin Trials 10,407-415[CrossRef][ISI][Medline]
  30. Rosner, B (1990) Fundamentals of Biostatistics 3rd ed. ,327 PWS-Kent Publishing Boston.
  31. March, L, Irwig, L, Schwarz, J, et al (1994) N of 1 trials comparing a non-steroidal anti-inflammatory drug with paracetomol in osteoarthritis. BMJ 309,1041-1046[Abstract/Free Full Text]
  32. Irwig, L, Glasziou, P, March, L (1995) The ethics of n of 1 trials. Lancet 345,469



This article has been cited by other articles:


Home page
 BMJHome page
G. E. Vist, K. B. Hagen, P J Devereaux, D. Bryant, D. T. Kristoffersen, and A. D. Oxman
Systematic review to determine whether participation in a trial influences outcome
BMJ, May 21, 2005; 330(7501): 1175.
[Abstract] [Full Text] [PDF]

Home page
 Eur Respir JHome page
J.R. Curtis and D.L. Patrick
The assessment of health status among patients with COPD
Eur. Respir. J., June 1, 2003; 21(41_suppl): 36S - 45s.
[Abstract] [Full Text] [PDF]

Home page
 Am. J. Respir. Crit. Care Med.Home page
R. A. PAUWELS, A. S. BUIST, P. M. A. CALVERLEY, C. R. JENKINS, and S. S. HURD
Global Strategy for the Diagnosis, Management, and Prevention of Chronic Obstructive Pulmonary Disease . NHLBI/WHO Global Initiative for Chronic Obstructive Lung Disease (GOLD) Workshop Summary
Am. J. Respir. Crit. Care Med., April 1, 2001; 163(5): 1256 - 1276.
[Full Text]

Home page
 ChestHome page
D. A. Mahler
How Should Health-Related Quality of Life Be Assessed in Patients With COPD?
Chest, February 1, 2000; 117(2_suppl): 54S - 57S.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF) Free
Right arrow Submit a response
Right arrow Alert me when this article is cited
Right arrow Alert me when eLetters are posted
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Article Archive
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via ISI Web of Science (19)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Mahon, J. L.
Right arrow Articles by Donner, A.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Mahon, J. L.
Right arrow Articles by Donner, A.

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE