LETTERS TO THE EDITOR Pill hoarding and hypercompliance To the Editor:
tient was able to surreptitiously sequester a supply of an
Two recent articles in CLINICAL PHARMACOLOGY AND THERAPEUTICS have highlighted problems inherent in the use
investigational drug for his own purposes because of the standard clinical trial method by which the drug was dispensed and accounted for. Pill hoarding may constitute another source
of pill counts as a measure of compliance with prescribed therapy during clinical trials.'.2 One study' identified a subgroup of "hypercompliers" whose pill counts indicated
of inaccuracyof entirely uncertain magnitudein the conduct of clinical trials.
James F. Burris, MD William J. Mroczek, MD Cardiovascular Center of Northern Virginia 4600 King Street, Suite 4A Alexandria, Virginia 22302
more than the prescribed number of doses had been taken but whose clinical response was less satisfactory than "eucom-
pliers." As pointed out by that study's authors, hypercompliance is possible because trial subjects are prescribed an excess of pills at each visit to permit modest postponement of visits when necessary; the excess can also serve as a check for "pill dumping." The authors suggested ingestion of extra doses, sharing of medications with nonstudy patients, inad-
vertent loss of pills, or deliberate discarding of medication as possible explanations for the observed hypercompliance and concluded that pill dumping was the most likely explanation. We have observed a case that suggests pill hoarding may be another source of apparent hypercompliance. Case report. F. W. U. was a 50-year-old white male businessman who participated in two clinical trials at our center before commencing a long-term study of a r3-adrenergic antagonist in August 1983.* The patient kept meticulous records of home blood pressures and subjective symptoms. He liked the study drug, which controlled his hypertension, had min-
imal side effects, and suppressed his mild stutter. At the conclusion of the study in November 1987, the patient was asked to return all unused study medication and was placed on a marketed angiotensin converting enzyme inhibitor. Approximately a year later, he mentioned that he occasionally used leftover study medication to suppress his stutter when he had an important business presentation to make. He was again asked to return any unused study medication, but he reported that he had none left. Almost a year after that, he developed symptomatic premature atrial contractions and reported that the study medication suppressed his palpitations; he went on to declare he "really was out of' the study medication this time. Review of study drug dispensing records revealed that pill counts were consistently correct or slightly hypercompliant.
Discussion. Study patients, particularly those who have participated in several trials, may become relatively sophisticated about clinical trial methods. As Rudd et al.' pointed out there are several clues that pill counts are being made even without explicit announcement of the fact. In the case reported here, an intelligent and medically sophisticated pa*The study in which the described patient was participating was sponsored by Lorex Pharmaceuticals, Skokie, Ill. However, there was no external support for this particular report.
References Pallar T, Kumar S, Tindall H, Feely M. Time to stop counting the tablets? CLIN PHARMACOL THER 1989;46:163-8.
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 .
Automated record linkage To the Editor: I would like to correct and comment on some statements in Dr. Shapiro's description of our Kaiser Permanente program to screen drugs for possible carcinogenic effects, which
appeared on pages 377 to 379 of his article, "The role of automated record linkage in the postmarketing surveillance of drug safety: A critique." Article: "For all statistically significant associations, the medical records of the cancer cases were reviewed." Correction: For most statistically significant associations the medical records of the cancer cases and a sample of users of the drug in whom cancer did not develop were reviewed. Article: ". . there were no data on the quantity of cigarettes smoked."
Correction: Data on the quantity of cigarettes smoked were available but were not considered in this study. Article: ". . diagnoses were validated in some instances." Correction: Diagnoses were validated in all instances by .
manual review of medical records.
Article: "Also, because many of the cases developed after several years, there was a limited capability to evaluate long latent intervals."
Comment: As follow-up of the drug cohort continues, longer latency periods can be studied. The limited capability in the study under consideration was not an inherent fault of record linkage. Article: "In the general analysis, coherence could hardly
547
tient was able to surreptitiously sequester a supply of an
Two recent articles in CLINICAL PHARMACOLOGY AND THERAPEUTICS have highlighted problems inherent in the use
investigational drug for his own purposes because of the standard clinical trial method by which the drug was dispensed and accounted for. Pill hoarding may constitute another source
of pill counts as a measure of compliance with prescribed therapy during clinical trials.'.2 One study' identified a subgroup of "hypercompliers" whose pill counts indicated
of inaccuracyof entirely uncertain magnitudein the conduct of clinical trials.
James F. Burris, MD William J. Mroczek, MD Cardiovascular Center of Northern Virginia 4600 King Street, Suite 4A Alexandria, Virginia 22302
more than the prescribed number of doses had been taken but whose clinical response was less satisfactory than "eucom-
pliers." As pointed out by that study's authors, hypercompliance is possible because trial subjects are prescribed an excess of pills at each visit to permit modest postponement of visits when necessary; the excess can also serve as a check for "pill dumping." The authors suggested ingestion of extra doses, sharing of medications with nonstudy patients, inad-
vertent loss of pills, or deliberate discarding of medication as possible explanations for the observed hypercompliance and concluded that pill dumping was the most likely explanation. We have observed a case that suggests pill hoarding may be another source of apparent hypercompliance. Case report. F. W. U. was a 50-year-old white male businessman who participated in two clinical trials at our center before commencing a long-term study of a r3-adrenergic antagonist in August 1983.* The patient kept meticulous records of home blood pressures and subjective symptoms. He liked the study drug, which controlled his hypertension, had min-
imal side effects, and suppressed his mild stutter. At the conclusion of the study in November 1987, the patient was asked to return all unused study medication and was placed on a marketed angiotensin converting enzyme inhibitor. Approximately a year later, he mentioned that he occasionally used leftover study medication to suppress his stutter when he had an important business presentation to make. He was again asked to return any unused study medication, but he reported that he had none left. Almost a year after that, he developed symptomatic premature atrial contractions and reported that the study medication suppressed his palpitations; he went on to declare he "really was out of' the study medication this time. Review of study drug dispensing records revealed that pill counts were consistently correct or slightly hypercompliant.
Discussion. Study patients, particularly those who have participated in several trials, may become relatively sophisticated about clinical trial methods. As Rudd et al.' pointed out there are several clues that pill counts are being made even without explicit announcement of the fact. In the case reported here, an intelligent and medically sophisticated pa*The study in which the described patient was participating was sponsored by Lorex Pharmaceuticals, Skokie, Ill. However, there was no external support for this particular report.
References Pallar T, Kumar S, Tindall H, Feely M. Time to stop counting the tablets? CLIN PHARMACOL THER 1989;46:163-8.
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 .
Automated record linkage To the Editor: I would like to correct and comment on some statements in Dr. Shapiro's description of our Kaiser Permanente program to screen drugs for possible carcinogenic effects, which
appeared on pages 377 to 379 of his article, "The role of automated record linkage in the postmarketing surveillance of drug safety: A critique." Article: "For all statistically significant associations, the medical records of the cancer cases were reviewed." Correction: For most statistically significant associations the medical records of the cancer cases and a sample of users of the drug in whom cancer did not develop were reviewed. Article: ". . there were no data on the quantity of cigarettes smoked."
Correction: Data on the quantity of cigarettes smoked were available but were not considered in this study. Article: ". . diagnoses were validated in some instances." Correction: Diagnoses were validated in all instances by .
manual review of medical records.
Article: "Also, because many of the cases developed after several years, there was a limited capability to evaluate long latent intervals."
Comment: As follow-up of the drug cohort continues, longer latency periods can be studied. The limited capability in the study under consideration was not an inherent fault of record linkage. Article: "In the general analysis, coherence could hardly
547
