Trends in Compliance With Bronchodilator Inhaler Use Between Follow-up Visits in a Clinical Trial

Trends in Compliance With Bronchodilator Inhaler Use Between Follow-up Visits in a Clinical Trial

Trends in Compliance With Bronchodilator Inhaler Use Between Follow-up Visits in a Clinical Trial* Michael S. Simrrwns; Mitchell A Nides, PhD; Cynthia...

3MB Sizes 0 Downloads 3 Views

Trends in Compliance With Bronchodilator Inhaler Use Between Follow-up Visits in a Clinical Trial* Michael S. Simrrwns; Mitchell A Nides, PhD; Cynthia S. Rand, PhD; Robert A Wise, MD; and Donald P. Tashkin, MD Study objective: To assess objectively measured, longterm trends in compliance with physician-prescribed metered-dose inhaler (MDI) use during a clinical trial. Design: A prospective study. Setting: The Lung Health Study, a 5-year clinical trial to determine the effect of special intervention with an intensive smoking cessation program and bronchodilator therapy in cigarette smokers 35 to 60 years of age with minimal to moderate airflow limitation due to COPD. Participants: Two hundred thirty-one participants who were issued an MDI with an attached Nebulizer Chronolog (NC) (Forefront Technologies Inc; Lakewood, Colo) which electronically records the date and time of each MDI actuation. One hundred two participants were not informed of the recording capabilities of the attached NC, while 129 participants were aware of the NC's monitoring function. Intervention: Following an initial 12-week period of counseling, participants retumed to the clinic every 4 months. Measurements and results: Analysis of the data from the NC collected over a period of 2 years indicates that

clinical trials to determine the efficacy of new treatments or to test novel applications of established treatments depend not only on a rigorous, carefully executed protocol, but also on the full cooperation of the trial subjects. Participants are responsible for daily self-administration of the study treatment at the prescribed intervals and for reporting of adverse events. It is well known that patient compliance with prescribed medication regimens is frequently less than optimal. 1 Poor compliance can affect the outcome of long-term clinical trials, possibly leading to inaccurate assessment of the treatment under investigation. However, self-report, the most commonly used method of assessing compliance, is frequently misleading, even *From the UCLA School of Medicine, Los Angeles (Mr. Simmons and Drs. Nides and Tashkin); and the Johns Ropkins University School of Medicine, Baltimore (Drs. Rand and Wise). This research is su_pported by contracts NOl HR 46022 and NOl HR 46016 from tfie Division of Lung Diseases, National Heart, Lung, and Blood Institute, US Public Health Service, Bethesda, Md, and by Boehrin_ger Ingelheim Pharmaceuticals, Inc. Manuscript received july 6, I995; revision acce12ted September 25. Reprint requests: Dr. Simmons, UCLA School of Medicine, Box 957042, Los Angeles, CA 90095-7042

compliance with the prescribed medication regimen was best immediately following each follow-up visit and gradually declined during the interval between follow-up visits. The level of compliance after each visit was lower for each successive follow-up. These trends could not be observed from self-report or weighing the medication canisters at follow-up visits. The participants who were informed of the NC's function and who were provided with detailed feedback about their inhaler use generally showed better compliance. (CHEST 1996; 109:963-68)

ANOVA=analysis of variance; JHU=Johns Hopkins University; MDI=metered-dose inhaler; NC=Nebulizer Chronolog; Sl=special intervention; UC:usual care; UCLA= University of California, Los Angeles

Key words: chronic obstructive pulmonary disease (CO PD); compliance; electronic medication monitoring; ipratropium bromide; metered-dose inhaler

among honest, well-intentionedstudysubjects. 2•3 Some objective methods of monitoring compliance, such as pill counts and inhaler canister weighings, are not only frequently inaccurate but also are not sensitive to trends in compliance over time and do not indicate whether the medication was taken at the prescribed times during the day. 3-6 Using a Nebulizer Chronolog (NC) (Forefront Technologies Inc; Lakewood, Colo) 2•7 attached to metered-dose inhalers (MDis), we electronically recorded each use of inhaled medication in a 5-year clinical trial (the Lung Health Study) in order to examine the patterns of use of the medication and objectively measure compliance with the prescribed medication protocol over the entire duration of the trial. Each individual in a subgroup of the study participants was provided with feedback of his/her actual inhaler usage data to permit us to examine the effect of such detailed feedback on deviations from the prescribed medication protocol as compared to a control group. Data collected over the first 24 months of the clinical trial are included in this report. Because of the CHEST I 109 I 4 I APRIL, 1996

963

large size of the study sample and the long period of data collection, this study provides a unique opportunity to investigate objectively measured patterns of medication usage in a clinical trial. The aims of the current study were to detennine (1) whether and to what extent compliance drops off over the interval between study visits, (2) whether follow-up visits improve compliance, (3) whether follow-up visits later in the study are more or less effective in improving compliance, and (4) whether any such changes in compliance are affected by feedback on the participant's inhaler use. MATERIALS AND METHODS

The Lung Health Study

The Lung Health Study, sponsored by the National Heart, Lung, and Blood Institute of the National Institutes of Health, is a 5-year clinical trial to determine the effect of special inteiVention with an intensive smoking cessation program and bronchodilator therapy in cigarette smokers 35 to 60 years of age with minimal to moderate airflow limitation due to COPD. 8·9 The subject population and study procedures have been described previously. 8 Participants were randomly assigned to either of two groups: "usual care" (UC) or "special inteiVention" (SI). UC subjects (n= 1,964) were contacted annually for repeated pulmonary function testing and determination of smoking status. SI subjects (n=3,923) were entered into an intensive group smoking cessation program with frequent contacts and follow-up visits to the study clinic every 4 months. At each follow-up visit, a questionnaire was administered that included questions about smoking status and inhaler use, and the inhaler canisters returned by the participants were weighed to determine the amount of medication that was used. Pulmonary function testing was repeated annually. SI subjects were also given an MDI containing either ipratropium bromide, an anticholinergic bronchodilator, or placebo by double-blind random assignment in order to evaluate the effect of such medication on the progression of COPD. They were instructed to use the MDI two inhalations three times per day. Nebulizer Chronolog Study

Two of the ten centers, the University of California, Los Angeles (UCLA) and Johns Hopkins University (JHU), participated in this ancillary study of inhaler compliance. A total of 1,117 participants were enrolled in the Lung Health Study at these two centers, 500 participants at UCLA (167 UC and333 SI participants), and 617 participants at JHU (206 UC and 411 SI participants). Two hundred forty-one consecutively enrolled SI participants at both centers (91 at UCLA and 150 at JHU ) received an inhaler with an attached NC, which electronically records the date and time of each MDI actuation. Ten participants were subsequently dropped from the analysis because they had no usable NC readings throughout the study due to technical problems, primarily battery failure and other electronic problems, and missed visits (three at UCLA and seven at JHU), leaving 231 participants in the study. One hundred two of these participants (46 at UCLA and 56 at JHU) were not informed of the date- and time-recording capabilities of the attached NC ("control" group), although they were aware that the NC would be monitoring the total amount of medication used. One hundred twenty-nine participants (42 at UCLA and 87 at JHU) were aware of the NC's exact monitoring function ("feedback" group). In the feedback group, the readings of actuation dates and times from the NC were used for feedback during the first and tenth weeks following their smoking cessation group's "quit" date and at each 964

4-month follow-up in an effort to enhance compliance with the prescribed medication regimen. Assignment to the NC study groups (control or feedback) was made by smoking cessation class; an entire class was assigned to either the control or feedback condition in order to avoid inadvertent unblinding of participants through interactions with classmates. Classes were first assigned to the control condition until all control participants were enrolled, with all subsequent enrollments to the feedback condition. This article reports on the first 2 years of data collection. The subjects were instructed to use their inhalers three times each day, taking two inhalations each time 1 to 5 min apart. To evaluate the pattern of inhaler use, all NC actuations were divided into daily sets of use, which is a measure of the number of separate times that the inhaler is used each day, with each day considered to begin at 4 AM to account for late-night inhaler use before going to bed. Starting with the first inhaler actuation, any actuation that occurs within a 3-h period is counted as a part of the same set. Mean sets per day for each 2-week inteiVal following initial issuance of the NC and following each follow-up visit were calculated. In both the control and feedback groups, changes in patterns of use following each follow-up visit were evaluated by changes in the mean sets per day. The difference between the first and last 2-week period of each follow-up inteiVal was used to assess the change during the intervisit inteiVal, while the effect of the follow-up visit on compliance was evaluated within and between both groups by comparing the compliance during the last 2-week period before the follow-up visit \vith compliance during the first 2-week period after the follow-up visit. Statistical Analysis

Unpaired t tests were used to compare the means of the control and feedback groups at each comparable follow-up period. Comparisons of the means of 2-week inteiVals within groups were made using paired Student's t tests. In the subset of participants whoreturned to the clinic and had usable data for every follow-up visit, repeated measures analysis of variance (ANOVA) was used to compare the means of the control and feedback groups (betweensubject effect), the means of all of the follow-up periods (withinsubject effect), and the interaction between these effects. Chisquare tests were used to compare the feedback and control groups \vith respect to the number of participants whose compliance increased or decreased after each follow-up visit from the compliance measured during the preceding inteiVal. Tests with a p value less than 0.05 were considered to be statistically significant. All statistical analyses were performed using a software system (SAS). JO RESULTS

The 241 participants who were issued an NC were seen in 1,446 follow-up visits during the 24-month study period. Of those, 309 follow-up visits (21.4%) were excluded from analysis due to technical difficulties reading the data from the NC or failure of participants to return the NC to the clinic for reading. Of the 1,137 remaining NC readings, 35 (2.4%) follow-up visits for which the interval between visits differed substantially from the interval dictated by the study protocol (<30 days or >40 weeks) were also excluded. A similar proportion of follow-up visits were excluded from analysis for both the feedback and uninformed groups. Ten participants had no usable NC readings, leaving 231 participants eligible for analysis, 102 participants in the control group and 129 in the feedback group. Eighty-seven participants who had data availClinical Investigations

CONTROL SUBJECTS

Table 1-Mean (±SD) Inhaler Use, as Number of Sets of Actuations per Day, for Each Follow-up Period (Perfect Compliance=3.0)

... 2.0

Group

Follow-up Period, months

Control

4 8 12 16 20 24

1.60:±:0.83 (85) 1 1.31:±:0.89 (84) 1.29:±:0.91 (78) 1.27:±:0.92 (73) 1.22:±:0.97 (69) 1.16:±:0.95 (75)

2 . 5 , - - - - - - - - - - -- - - - - - - - ,

4 MONTH FOLLOW -UP

c

Cl

Feedback 1.93:±:0.69 (107)1 1.76:±:0.83 (112) 1.74:±:0.89 (104) 1.70:±:0.89 (104) 1.56:±:0.87 (101) 1.65:±:0.89 (101 )

p Value* 0.0035 0.0003 0.0007 0.0018 0.0190 0.0006

....

~ WONTH

~~~;~~~~~~~~~~~=~

"... ---··-··-~~::::;::~\~·c:,."

= z

1.0

c

I

0.5

0 .0 '-----L--'---.J..._----L_ 2

*p value for unpaired t test comparison of control vs feedback groups. 1Number in parentheses represents the number of subjects in whom data were evaluable for the corresponding follow-up period.

able for analysis from all visits were included in the repeated measures ANOVAs comparing the compliance and intraperiod decline between groups. Sixtynine participants had complete data for the comparisons of postfollow-up increases in compliance, which required the 28-month NC reading in order to assess the effect of the 24-month follow-up visit. Twentythree of the 231 participants used in the analysis (10%) were permanently unavailable for follow-up before their 24-month follow-up visit, 14 in the feedback group (10.9%), and 9 in"the control group (8.8%). The proportions of subjects in each group who were unavailable for follow-up were not statistically significantly different (x 2 test; p=0.61). Comparison of the baseline characteristics between the feedback and control groups revealed that the control group was slightly older than the feedback group (50.3 vs 48.4 years, unpaired Student's t test; p=0.027), had a slightly lower percent of predicted FEV1 (71.5% vs 73.9%, unpaired Student's t test; p=0.031), and was slightly more responsive to a bronchodilator (16.7% vs 8.3% responding, x2 test; p<0.001). There were no statistically significant differences between the groups in gender, frequency of physician-diagnosed pulmonary disease (asthma, bronchitis, and emphysema), race, education, or baseline cigarette smoking amount. The mean numbers of sets per day are shown in Table 1 for each subject group (control/feedback) and each 4-month follow-up period. Data for each follow-up period represent inhaler usage during the previous 4 months. Inhaler use was significantly higher in the feedback group than in the control group during all follow-up periods considered individually (unpaired t test; p=0.019 to 0.0003) or simultaneously among the subset with complete data for all visits (ANOVA; p=0.0002). No significant interaction was noted between group and follow-up period. The means of the 4-, 8-, and 16-month follow-up periods were found to differ significantly from the immediately subsequent periods (ANOVA; p=0.037 to 0.0001 ). Compliance during the first 4-rrwnth follow-up pe-

--

1 1.5

4

6

8

10

_ J _ _ . l _ _ _ _ . L_

12

16

14

_J__...J..____>~__:j

24

22

20

18

WEEKS

MO NTH OF FOLLOW· UP VISIT :

-----

16

12

- - ---

20

- ---

24

FEEDBACK SUBJECTS 2.5 , - - - - - - - - - - - - - - - - - - - - ,

... 2.0

~

..

1 1.5

I;;

.. 1.0

z

;

0.5 0 .0 ' - - - - - L - - ' - - ' - - _ _ l ._

2

4

6

MO NTH OF FOLLOW-UP VISIT:

8

10

_ J _ _ . l _ _ _ _ . L_

12

16

14

_L..._...J..___._

20

18

____J

24

22

WEEKS 12

16

- - -·-

20

----

24

FIGURE l. The mean number of sets of inhaler usage per day for each 2-week interval following each follow-up visit are shown. The lines are labeled for the follow-up visit at which the data stored in the NC were read. Data for each subject group (control/feedback) are shown separately. Mean inhaler usage was always below the prescribed three sets per day. Compliance improved immediately following each follow-up visit and declined until the next follow-up visit. Compliance immediately after each follow-up visit declined with each successive visit; the decrease in compliance between visits was most evident during the earlier follow-up periods. The feedback group maintained a higher overall level of inhaler use throughout all follow-up intervals.

riod (mean sets per day) was significantly higher for the subjects who were still participating in the study after 24 months than for those who had dropped out (failed to return for their 24-month follow-up visit). Among the feedback subjects, means for the first 4-rrwnth follow-up period were 1.99 and 1.40 sets per day for the continuing and dropped subgroups, respectively (unpaired Student's t test; p=0.004). The control subjects had means of 1.67 and 0.87 sets per day for the continuing and dropped subgroups, respectively (unpaired Student's t test; p=0.015). The mean number of sets per day for each 2-week interval following initial issuance of the NC and each follow-up visit is shown in Figure 1 for control and feedback subjects, respectively. The means for 20 to 24 weeks after each follow-up visit represent the subset of participants who were late returning for their follow-up CHEST / 109 / 4 / APRIL, 1996

965

Table 2-Change in Inhaler Usage From First 2-Week Interval After Follow-up Visit to Last 2-Week Interval Before the Next Follow-up Visit* Follow-up Period, months

Control

p Value 1

4 8 12 16 20 24

-0.68:'::0.76 (85)1 -0.42:'::0.76 (84) -0.19:'::0.66 (78) -0.39:'::0.70 (73) -0.23:'::0.61 (69) -0.20:'::0.68 (75)

0.0001 0.0001 0.0140 0.0001 0.0027 0.0120

Group Feedback

p Value 1

-0.60:'::0.88 (107)1 0.0001 -0.32:!:0.065 (112) 0.0001 -0.29:'::0.63 (102) 0.0001 -0.34:'::0.74 (104) 0.0001 -0.29:'::0.73 (101) 0.0002 -0.14:'::0.68 (101) 0.0450

*Mean (:'::SD) (N) difference in sets of actuations per day. 1p value for paired t test comparison of first 2-week interval vs last 2-week interval. INumber in parentheses represents the number of subjects in whom data were evaluable for the corresponding follow-up period.

visit. Inhaler use was closest to the prescribed three sets per day immediately after issuance of the NC (beginning of the 4-month line), the feedback subjects demonstrating greater initial compliance than the control subjects. For both groups, inhaler use rapidly declined, the greatest decline occurring during the first 8 weeks after the NC was issued. The control group demonstrated a greater decline in inhaler use throughout the first 4-month follow-up period than did the feedback group. Both groups' compliance improved immediately following the 4-month follow-up visit (beginning of the 8-month line), but neither group's inhaler use returned to the level of use demonstrated at the beginning of the study. Feedback subjects showed more improvement in inhaler use following the 4-month visit than did the control group. The level of compliance immediately after each follow-up visit was lower in each successive visit in both groups, the decrease in compliance between visits tending to be most evident during the earlier follow-up periods. While the declines of the two groups over the first 16 to 18 weeks of all but the first 4-month interval were similar, the feedback group maintained a higher overalllevel of inhaler use throughout all follow-up intervals. Similar results were obtained when the data from each center were examined separately. Changes in inhaler usage from the first 2-week period after each follow-up visit to the last 2-week period before the next follow-up visit are shown in Table 2. Differences between the first and last 2-week intervals were statistically significant at all follow-up periods for both groups (paired Student's t test; p=0.045 to 0.0001). No significant differences were noted between groups, and no significant interaction was found between group and follow-up period (ANOVA). The differences in the mean sets per day between the 2-week intervals immediately before and immediately after each follow-up visit are shown in Table 3 for each subject group and follow-up visit. Significant in966

creases in mean sets per day were found immediately following all follow-up visits for both groups (paired Student's t test; p=0.028 to 0.0001). No significant differences were found between the groups. Additional analyses were performed using mean compliant days, with each subject's daily inhaler usage defmed as either compliant or noncompliant. Compliant day was defined as any day that contained two to four sets with at least two of the sets consisting of two or three inhalations. Results were similar to those obtained with the analysis of mean sets per day. DISCUSSION

Using the NC to monitor MDI usage in a clinical trial, we found that compliance with the physicianprescribed bronchodilator MDI protocol varied considerably over the 24-month observation period. Compliance decreased during the 4-month intervals between follow-up visits, but increased immediately after each of the visits. This pattern was most pronounced early in the study, with participants whoreceived feedback about their actual MDI usage maintaining a higher level of compliance throughout the trial. These findings may be affected by biases introduced by factors related to the long duration of the trial, including missed or delayed visits and dropouts over the course of the study. The decrease in compliance evident at later times (from 18 to 24 weeks) in each follow-up period in Figure 1 may be due in part to the fact that the participants who returned for their follow-up clinic visits after the scheduled time of the visit tended to be the least compliant participants. Thus, data in the later weeks of each follow-up period are collected from a higher proportion of those who are undercompliant. Changes in the pattern of compliance during the later follow-up periods could also be affected by the progressive loss of the least compliant participants, as Table 3-Change in Inhaler Usage From the Last 2-Week Interval Before the Indicated Follow-up Visit to the First 2-Week Interval After the Indicated Follow-up Visit* Follow-up Period, months 4

8 12 16 20 24

Group Control

p Value 1

Feedback

p Value 1

0.19:'::0.58 (72)1 0.17:'::0.44 (69) 0.15:'::0.50 (63) 0.23:'::0.57 (62) 0.17:'::0.41 (63) 0.20:'::0.60 (63)

0.0065 0.0022 0.0180 0.0026 0.0016 0.0091

0.23:'::0.73 (95)1 0.23:'::0.56 (91) 0.24:'::0.67 (89) 0.18:'::0.65 (88) 0.19:'::0.77 (85) 0.24:'::0.53 (78)

0.0033 0.0001 0.0010 0.0120 0.0280 0.0001

*Mean (:'::SD) (N) difference in sets of actuations per day. 1p value for paired t test comparison of last 2-week interval before visit vs first 2-week interval after visit. INumber in parentheses represents the number of subjects in whom data were evaluable for the corresponding follow-up period. Clinical Investigations

suggested by the significantly higher compliance during the initial 4-month period of those who remained in the study for the full24 months as compared to those who dropped out. This bias could cause compliance later in the study to appear higher than that which would have been found had the less-compliant participants not dropped out of the NC study. Similarly, this bias would cause compliance to appear to be steadier, with a less-steep decline, throughout the follow-up intervals than if the less-compliant participants had remained in the study. While microelectronic monitoring of the type used in this study can provide an accurate record of the date and time of each actuation of the MDI, it cannot determine how much of the medication was actually inhaled. Other studies have reported patterns of behavior that have been interpreted as a deliberate attempt to deceive the investigators by intentionally emp~n~ containers of study medication prior to clinic visits. ·6•1 When a regular and appropriate pattern of inhaler actuations is recorded, however, it seems unlikely that a subject would fail to inhale the medication after having taken the trouble to follow the study protocol. Some subjects, however, may fall somewhere between these two extremes. The finding in the present study that compliance decreases during the interval between follow-up visits in a clinical trial is consistent with previous reports. Using an electronic pill monitor, Cramer et al 12 found that in a group of 20 patients with epilepsy, the average compliance with medication prescribed for seizure control decreased durin§ the interval between clinic visits. Wanamaker et al 1 observed improvements in drug levels and seizure status when the interval between clinic visits was shortened in a group of 30 patients on stable regimens of antiepileptic drugs, implying that compliance had decreased during the longer interval between clinic visits. The present report extends these investigations of the day-to-day patterns of compliance to a much longer clinical trial (2 years). In contrast to the steep decline observed during the first 4 months of the present trial, the rate of decline diminished during subsequent follow-up intervals, especially for the feedback group. In a report based on data from only the first 4-month follow-up interval of the present study, Nides et al 11 found that compliance was significantly higher among those subjects who were aware that their MDI use was being electronically monitored and received feedback based on actual MDI usage records between the regular follow-up visits compared to the uninformed control group. This article confirms the persistence of this difference during subsequent follow-up intervals. In addition, the present report examines the relative effect of the follow-up visit itself in the subjects given feedback compared to subjects who were uninformed

of the exact monitoring nature of the NC. The relatively higher compliance of the feedback group after the initial follow-up visit could be due at least in part to corrections in their behavior induced by the feedback that they received. Improvements in behavior may then be maintained at least partially for the duration of the study. An alternative explanation for these findings could be that after the initial follow-up period, the feedback group's compliance was improved not because of a direct effect of being informed of their individual MDI usage patterns at the follow-up visit, but rather because they were aware of the full monitoring nature of the NC. Because compliance increased significantly following all visits for the control group and the feedback group, the follow-up visits appear to have encouraged the participants to adhere to the prescribed inhaler protocol whether or not they knew the full extent of the NC's monitoring capability. It is possible that contact between follow-up visits by telephone or mailed reminders might also serve to improve compliance, but this was not tested by the present study. The findings from the present study may have implications for the interpretation of data from long-term clinical trials. It is not possible to limit analysis to only fully compliant individuals in a lengthy study if compliance decreases eventually in nearly all subjects. Partial compliance then assumes importance. Investigators must consider the impact of varying degrees of compliance throughout a clinical trial on the outcome. If compliance decreases during the interval between clinic visits, how does this affect the interpretation of outcome variables measured only at the clinic visits or self-assessed by the subjects throughout this period of declining drug use? The temporary increases in drug use just prior to the clinic visit reported by some investigators 12.l4 further complicates the analysis, in extreme cases effectively reducing a planned long-term study to a series of short-term studies. The gradual decrease in objectively monitored compliance over the course of a clinical trial lasting many months or years, even in the face of direct feedback with precise compliance monitoring data, is an important finding of this study. Despite strenuous attempts to maximize compliance in a long-term clinical trial, therefore, it should be anticipated that compliance will be suboptimal and decline somewhat over time. Frequent feedback based on objective compliance monitoring can improve compliance but may not always be feasible . The decline in compliance between clinic visits may be reduced by more frequent visits to the clinic, but this may not always be practical. However, the subject's awareness that medication use is being monitored may in itself be sufficient to improve compliance. Our findings also suggest that compliance monitoring may be particularly important in lengthy CHEST I 109 I 4 I APRIL, 1996

967

clinical trials, since a general decrease in compliance can occur later in the study that is not as likely to be reversed by a clinic visit. In addition, since both the feedback and control groups demonstrated similar relative declines after the first 4-month. follow-up visit, it may be particularly important to establish good compliance behavior early in the course of clinical treatment or a clinical trial. However, the high percentage of missing data in our study, due to missed clinic visits and technical difficulties with reading the NC, suggests that researchers may not want to depend entirely on electronic monitors for compliance data. These observations emphasize the need for developing improved microelectronic devices and other strategies for enhancing and monitoring compliance. ACKNOWLEDGMENTS: The authors wish to thank Anne H. Coulson for her valuable assistance in developing the study protocol and Kathleen Weeks for collecting and editing the NC data at the JHU center.

4

5

6 7

8

9

10 REFERENCES

1 Cramer JA. Overview of methods to measure and enhance patient compliance. In: Cramer JA, Spilker B, eds. Patient compliance in medical practice and clinical trials. New York: Raven Press, 1991; 3-10 2 Spector SL, Kinsman R, Mawhinney H, et al. Compliance of patients with asthma with an experimental aerosolized medication: implications for controlled clinical trials. J Allergy Clin Immunol 1986; 77:65-70 3 Waterhouse DM, Calzone KA, Mele C, et al. Adherence to oral

968

11

12 13

14

tamoxifen: a comparison of patient self-report, pill counts, and microelectronic monitoring. J Clin Oncol1993; 11:1189-97 Cramer JA, Mattson RH, Prevey ML, et al. How often is medication taken as prescribed?: a novel assessment technique. JAMA 1989; 261:3273-77 Rudd P, Byyny RL, Zachary V, et al. The natural history of medication compliance in a drug trial: limitations of pill counts. Clin Pharmacol Ther 1989; 46:169-76 Rand CS, Wise RA, Nides M, et al. Metered-dose inhaler adherence in a clinical trial. Am Rev Respir Dis 1992; 146:1559-64 Gong H, Simmons MS, Clark VA, et al. Metered-dose inhaler usage in subjects with asthma: comparison of Nebulizer Chronolog and daily diary recordings. J Allergy Clin Immunol 1988; 82:5-10 Anthonisen NR, Connett JE, Kiley JP, et al. Effects of smoking intervention and the use of an inhaled anticholinergic bronchodilator on the rate of decline of FEV1-the Lung Health Study. JAMA 1994; 272:1497-1505 Connett JE, Kusek JW, Bailey WC, et al. Design of the Lung Health Study: a randomized clinical trial of early intervention for chronic obstructive pulmonary disease. Control Clin Trials 1993; 14:3-19 SAS/Stat users guide, version 6. 4th ed (vol1 and 2). Cary, NC: SAS Institute, 1989 Nides MA, Tashl
Clinical Investigations