HOW
WELL
DO
PATIENTS
TAKE
ORAL
PENICILLIN?
A
COLLABORATIVE
STUDY
IN
PRIVATE
PRACTICE
Evan Charney, M.D., Rufus Bynum, M.D., Donald Eldredge, M.D., Donald Frank, M.D.,
James B. MacWhinney, M.D., Neal McNabb, M.D., Albert Schemer M.D.,
Edwin A. Sumpter, M.D, and Howard Iker, Ph.D.
Department of Pediatrics, University of Rochester School of Medicine and Dentistry and Strong Memorial Hospital, Rochester, New York
188
it is as important to know what patient has
the disease as what disease the patient has.
T
HERE have been a number of studies inthe recent past concerning patient
compliance with a variety of medical
regimens,’ and several in the past decade
concerned specifically with the taking of
penicillin either for therapy for
streptococ-cal pharyngitis or for rheumatic fever
prophylaxis.7 In one study with clinic
pa-tients only 18% of the children completed
the penicillin therapy as prescribed. Two
recent studies from private 67
(
pub-lished while the current study was under
way
)
have concluded that more of those pa-tients receive their full 10 days oftreat-ment. The authors of these latter studies
suggest that either the nature of the patient seen or the private practice relationship itself may account for their better results.
The present study was done to obtain the
following specific information:
1. What percent of children are taking
oral penicillin five days and nine days fol-lowing its prescription for either strepto-coccal pharyngitis or otitis media?
2. \Vhat characteristics of patient, physi-cian, or disease differentiate those who take
the medication from those who do not?
This study was a collaborative effort be-tween seven pediatricians in private prac-tice (in three separate practice groups) and
The Department of Pediatrics at the
Uni-versity of Rochester School of Medicine and Dentistry. All of the patients are from the private practices of the three groups in Rochester
(
Monroe County)
, New York.The first problem, of course, is to deter-mine whether or not the child has taken the penicillin without influencing his behavior by the study itself. In previous studies ci-ther the parents were asked directly wheth-en or not the child had taken the me(licine,
the number of pills were counted on a
home visit or in the physician’s office, or
the child’s urine was examined for the pres-ence of penicillin.-#{176} Although there are
some disadvantages to all of these
tech-niques, it was decided to use the urine
col-lection technique and overcme the
disad-vantage of not being able to detect missed doses in more than the day of collection by
obtaining specimens from sonic patients at
five days and some at nine days after onset of treatment.
SARCINA LUTEA METHOD FOR DETECTION OF PENICILLIN IN URINE
A preliminary evaluation of this method
was made prior to the start of the study to
determine its accuracy, reliability, and ap-plicability for use in a private office labora-tory. Grove and Randall’ state that as little as 0.005 units per milliliter o penicillin can be detected in the urine, and Bergman5
(Received January 16; revision accepted for publication March 1 1, 1967.)
Supported in part by Children’s Bureau Grant # 148, United States Department of Health, Education, and Welfare.
Funds for purchasing culture material and equipment supplied by Wyeth Laboratories, Philadelphia,
Pennsylvania.
The authors represent a group of practitioners and full-time staff from the University of Rochester Medical School engaged in a program of collaborative clinical research.
ADDRESS FOR REPRINTS: (E.G.) Strong Memorial Hospital, 260 Grittenden Boulevard, Rochester, New York 14620.
HOURS AFTER DOSE
in urine following single dose of Pen-V, 125 mg. The
children are ages 2 to 12 years. ARTICLES
FIG. 1. Penicillin
noted that peilicillin was still detectable 16
hours after the last dose. The following
technique was used.
A 3 in. blank sterile filter paper disc* was
dipped in the urine specimen to be tested
and then placed on a sheep’s blood agar
plate that had been completely streaked
with an undiluted broth culture
(
at least 24 hours old)
of S. lutea. The plate wasincu-bated for at least 18 hours at 37#{176}C,at
which time a zone of inhibition was clearly
evident arouiid the disc if penicillin was
present in concentration greater than 0.08
‘,i.g/ml. All of 103 urine specimens
contain-ing penicillin showed an inhibition zone
around the filter paper disc. A total of 98
urine specimens from patients who had no
penicillin for at least three days were tested
and 91 of these showed no inhibition zone.
Of the remaining seven cases, four patients
had taken other antibiotics and three had
taken a variety of medicines but no
anti-biotics. Thus, three of 98 specimens showed inhibition of S. lutea growth, presumably due to some factor other than antibiotic. In
these three cases the radius of the zone of
inhibition was less than 10 mm. We were
o Obtained from Baltimore Biological
Labora-tory, Baltimore, Maryland.
unable to differentiate penicillin from other
inhibitors with any consistency by the use
of penicillinase impregnated discs. In short,
the technique seems to have no false
nega-tives
(
penicillin taken but not evident inurine
)
and only an occasional false positivewhich showed less inhibition than that
caused by penicillin.
The technique has several other
advan-tages. Contamination of the urine specimen with a variety of gram positive and gram
negative organisms does not alter the zone
of inhibition. The presence of albumin, leti-kocytes, or erthrocytes in the urine does not
affect the test. The urine specimen can be
left unrefrigerated for at least 12 hours
(
we did not test any for a longer period)
or re-frigerated for at least 48 hours before any change in the inhibition zone is evident.The forceps used to hold the filter disc
needs to be rinsed thoroughly with tap
water between specimens or else sufficient penicillin may remain on it to give a false positive result for the next urine tested.
Finally, the technique was standardized for an oral dose of 200,000 units of potas-sium phenoxymethyl penicillin
(
PenicillinV-K). Figure 1 compares the radius of
TABLE I
FAKE It
Days after
renictuin
rrescrioea
ATE AT 5 AND 9DAYS 0 N 459 PATIENTS
i aee ± No
Take
. Dzsqua/-1
.
ifled
.
Total .
Patients
Five (lays 79 9 10 9 107
Nine days 155 37 84 76 35
‘Fake-penieilliii taken within 15 hours.
Intermediate ( ±)-some penicillin between H and
0 hours.
No take-no penicillin for at least 18 hours.
l)isqualified-urine specimen returned more than 24
hours after (late requested, patient seen in interim by
(loctor, or patient unable to keep appointment.
standardization the child was considered to
be a no-taker if there was a zone of
inhibi-tion of less than 5 mm around the filter
paper disc
(
no penicillin for at least 18hours
)
. He was considered a taker if there was a zone of greater than 10 mm of inhibi-tion(
some penicillin within past 15 hours), and an intermediary case if the inhibitionzone was between 5 and 10 mm. Some of
the children receiving liquid penicillin
un-doubtedly took less than full 5 ml per dose.
The results in Figure 1 are based on what
parents give their children when “1
tea-spoon” is prescribed, and the range of
peni-cillin levels might well be narrower if the
dose were more precise. Since the study
was to be done on penicillin given by
par-ents at home, it was felt that this curve
would better reflect actual practice than
one constructed from a precise 5 ml dose.
METHOD
A preliminary study was done with 96
patients to determine whether the method
of obtaining the urine specimen influenced the “take rate.” No statistically significant
difference was found when the urine was
collected in the doctor’s office
(
mother toldonly to bring child back for recheck
)
orwhen the mother returned with both a
urine specimen collected at home and a
questionnaire filled out.
The major portion of the study was then
conducted between May and June of 1965
and December 1965 and April 1966. All
children between 2 and 12 years of age were included in the study if oral penicillin was prescribed for either streptococcal pharyngitis
(
proved by culture)
or otitis media. The parent was given a prescription(
to be filled at their local pharmacy)
for ci-ther 150 ml of potassium phenoxymethyl penicillin(
Penicillin V-K)
or 30 tablets(
200,000 units each)
and instructed to take 1 teaspoon(
or 1 pill)
three times a day for the full 10 days. They were asked to returna urine specimen collected from the child
on either the fifth or ninth day of therapy
(
alternate patients)
. We made no specialcheck on whether the medicine was
actual-ly purchased. However, of a sample 28 families interviewed at home during a later phase of the study, all had the penicillin.
At the end of the study, but without
knowing the urine results, each practitioner completed an adjective check list
(
modifiedfrom Cough9
)
describing the mother inorder to obtain an appraisal of her
person-ality characteristics; he then predicted
whether or not the child had taken the pen-icillin.
RESULTS
A total of 459 children, age 1.5 to 14
years
(
median 5 years) ,
were begun in the study. The number of patients in each cat-egory is indicated in Table I. The familieswere largely middle and upper middle
class, mostly suburban and many in profes-sional occupations.
The possible influence of a variety of
factors on “take rate” was investigated.
Statistical analysis was one between the
“take” and “no take” groups using the
Chi-Square method. The children with
phar-yngitis completed the full course of
ther-apy somewhat more frequently than those
with otitis media (66% and 51%,
respec-tively). This difference approaches but
does not reach significance at the 5% level
(X2 2.92, p = .10). It was felt that the
difference may be a meaningful one,
‘FABLE II
Patients
Iota I gruiij
I’liaryiigitis aloite
Not ill
()titis media alone %lildly ill
(lf 2
p= .10
(lf 2
p= .90
%‘2=7 .18
df= 2
l= #{176}5
\I0TI1EIiS E.T1\1STF: OF EVE1I1TY 01’ I)ISESSE AT FIRST DocTon \ISIT
. S’ignifieant
: 1)egree (if Sererzt,j- . Take \o Tale Differenee
Not ill 14 (5 ) 13(48
\IiI(IIV ill 41 (6P ) 26 (39(
\Io(lerately or severely ill 94 (71 ) 38 (9’
Not iii 8 (73(,() 3 (7(;)
bIiI(lIyill 17 (7P ) 7 (‘29 )
Moderately or severely ill ‘37 (73 1 14 (‘27w)
6(38) 10(6)
4(56’) 19(44)
Moderately or severely ill 57 (70j 4 (30 )
disease separately as well as for the
com-biiied group. Significant differences (p =
< .05) for otitis media alone are marked with an asterisk0, for pharyngitis alone with a dagger.
Disease or Drug Factors
I. Otitis media or pharyngitis.
2. Degree of severity at onset, as
per-ceived by mother.#{176}
3. Degree of severity at onset, as per-ceiVe(l by doctor.
4. Duration of symptoms
(
as rated by mother).5. Type of ieiiicillin given
(
pill or liq-uid).6. Other medicines given concurrently or
not.
7. Throat culture taken or not (for otitis media).
Patient or Family Factors
1. Age of child. 2. Sex of child.
3. Number of children in family. 4. Position of child in family.
5. Intercurrent illness in other family
members (with and without regular medi-cine being given).
6. Educational level of parents.
7. Previous serious illness in child.
Doctor or Patient-Doctor Factors
1. Individual doctor.
2. Practice.
3. Number of years family cared for by that doctor.f
4. \Vhether patient seen by own doctor
or by partner.#{176}
5. Doctor’s description of mother’s
per-sonality h#{176}
f
Only four of these factors proved to
differ significantly between the “takers” and the “non-takers;” the mother’s perception of severity of disease at onset, whether or not the medicine was prescribed by the child’s
“usual doctor,” the number of years the
family had been cared for by that practice, and the mother’s personality characteristics. Specific data for the first three of these are
shown in Tables II-IV.
In general the mothers considered their
child’s illness more severe than the doctor
stan-Patients Prescribed by Take
Significant
No Take Differenee
Own physician
Total group ---
---Partner
105 (78) ‘39 (27%) .V 7.59
--- (1f’ I
37 (46) p < .01
---43 (54(’)
Own physician
Pharyngitis alone --
--- Partner
40 (74) 14 (26 ) X = 0.()6
--- (If = 1
9 (31 ) p= .80
20 (69 )
()titis media alone
Own physician 65 (72) 25 (28)
Partner 23 (45%) 28 (55’)
dardize the definition of “mild,”
“moder-ate,” or “severe” illness. It is of particular
interest that the children who took all the
penicillin were not symptomatic for any
longer time than those who did not
corn-plete therapy. Both “takers” and “non-tak-ers” were well, on the average, on the third day of treatment.
An attempt was made to assess the
importance of the mother’s personality
through use of an adjective check list
(
modified from Gough,i copies available onrequest
)
. Ten of the 62 adjectives showedsignificant differences between “take” and
“no take’ groups at the p < .05 level, two of
these at the p < .01 level. Table V lists these
10 adjectives. The remaining adjectives
were applied more or less evenly to both
groups. The “takers” are ascribed more
pos-itive adjectives than the “nontakers,” and
the reverse holds true for the negative ad-jectives. The intermediate and disqualified
groups in general are rated between two
extremes. Despite the fact that it was
known that the overall take rate was about
two-thirds, the doctors could not predict
this with better than chance accuracy. During the latter portion of the study
two first year medical students
inter-viewed 28 families between 1 and 3 weeks
after they had returned a urine specimen.
Costas Hercules and Norman Spack.
The students were unaware of whether the
child had taken the full penicillin course
prior to their interview. This number of
in-terviews is not large enough to warrant
de-tailed presentation but certain of the infor-mation obtained is Iertinent. The student
interviewed the mother in all cases and
asked to see the penicillin bottle at the end
of the interview in order to check the
amount of unused penicillin. Sixteen of the children were “takers,” 10 were
“non-tak-ers,” and 2 were “intermediate” by the
urine criterion. All of the mothers knew the
nature of the illness their child had had,
and all correctly identified the medicine
that had been prescribed and knew’ the
cor-rect dosage and intended duration of thera-py. In four cases the parents admitted that the child did not take all the penicillin and this finding was confirmed by the urine test.
The remaining six, however, claimed that
their child had used at least 4 of the 5 oz of
penicillin prescribed but there was no
peni-cillin present in the urine. One parent had more than 1 oz of penicillin remaining, but
the child’s urine contained penicillin. On
the basis of these findings it would seem
that the urine test is indeed a stricter
crite-non of take rate than a pill count, a fact
previously noted.5
As far as the students could determine
there were no more significant family crises or recent upsetting events in the
“non-tak-TABLE III
%Vo PRESCRIBES PENICILLIN?
9.09
(If =1
Take
116 (69)
29 (541;)
52 (80)
Less than 4
9 (47(;)
64 (62)
More than 4
Io(s3;)
39 (38w)
20(57w) TABLE IV
YEARS FAMILY UNDER CARE BY PHYSICIAN
Patients I’ears Under Care
Less than 4
lotal group
More than 4
Less than 4
Pliaryngitis alone
More than 4
Otitis media alone
Significant ‘so Take
Difference
52 (‘31’) .V2=3.6()
(If 1
25 (46) P= .06
13 (20 ) X=6.32
(If = I
= <#{149}0
X2=0.1()
(1f1
15(43)
ers” as opposed to the “takers.” Indeed, few
upsetting events were reported in either
group. None of the mothers seemed aware
of the purpose of the study, although most
knew that their child had been involved in
one. Before asking to see the penicillin bot-tle
(
the last question in the interview)
thestudents attempted to predict whether the
child had been given the full course of
medicine. They were unable to do so with
better than chance accuracy.
COMMENT
Only 56% of the children completed the prescribed 10-day course of penicillin ther-apy. Another 13% had taken some
penicil-lin but not in the past 15 hours. This is
certainly a better compliance rate than the
19% rate that has been reported from a
clinic population5 but considerably lower
than the 89% and 99% rate reported on
pa-tients with streptococcal pharyngitis from
other private practice settings.6’7 In one of
the private practice studies Leistyna and
Macauley gave each parent a printed
in-formation sheet on the hazards of strepto-coccal infection at the time the medicine was prescribed and this may well have em-phasized the necessity for complete therapy in the parent’s mind. They also asked their patients to return with the penicillin bottle at the end of therapy. If this will improve
patient compliance perhaps it ought to be
considered for use in routine practice!
The factors that do not distinguish the
“takers” from the “non-takers” are perhaps as interesting as the ones that do. The age
and sex of the child, the duration of his
symptoms, the doctor’s estimate of severity of the disease, the educational level of the
parents, and intercurrent illness in other
family members all seem unrelated to how well the medicine was taken. Streptococcal
phaiyngitis may be considered by the
fami-ly a more serious disease than otitis media
and this may account for its somewhat
bet-ter completion rate, although the children appeared equally ill with either disease.
The child who is thought by either the
mother or the doctor to be asymptomatic
but has an inflamed ear had less than a
50% chance of completing his full,
pre-scribed course. Probably the most
impres-sive correlations seem related to aspects of the patient-doctor relationship. If penicillin
was prescribed by a member of the
pedi-atric group other than the child’s “usual doe-tor”-in particular for otitis media-the pa-tient clearly was less likely to get the full
course. We did not specifically ask how
well the parent knew the other doctor, but
the physicians in each of the three groups
work closely toegether and rotate night
coverage so that some contact with all the
doctors in the group would be quite likely.
TABLE V
PERCENT OF PATIENTS ASSIGNED ADJECFIVE
.4djerlire T(ik(r’ Inter-
J)iquuli-- tuker., mediate tied
Responsible’ Organized’ Efficient’ Industrious’ Intelligent’ Clear thinkingt l.ogieiI’ Trusting’ :Iat,irc’ Unreliable’ Predicated “take
Predicated “no take
S:I (It; 70 75
3, :3-2 46 15
55 4 4t 41
55 16 53 4l
71 3t 67 31
41 ‘1(1 ei 16
15 O 1
59 In 46 44
30 10 17 47
7 1 S 11
64 61 6 47
4 35 9
Total patients rated I I I .56 t4
3-Difference between takers and non-takers significant: ‘p <03,
t p<.01
three group practices as well as all three
combined, which further strengthens the validity of the result.
Are mothers who are more likely to follow their doctor’s advice indeed more responsi-ble, reliable, and intelligent; or, does this
perception by the doctor reflect a warmer
and closer relationship with the family
which in turn leads to better compliance? Certainly both factors may operate togeth-er. It is of particular interest that the
pedia-tricians were unable to predict how well
the patients took their medication despite having differentiated the two groups by the adjective list. No one doctor was significant-ly better than any other at this prediction.
The disqualified patients do not differ
from those ‘ho completed the study in
type of disease, age of child, or severity of illness. They most resemble the
intermedi-ate group in the adjective description but
have somewhat fewer adjectives checked
per mother than either the “takers” or the
“non-takers,” perhaps reflecting less familiar-ity with the pediatrician.
We would draw several conclusions from
this study. Of most importance, too many
children do not receive their full course of
oral therapy. Second, the likelihood that a
child will receive his full course of
treat-ment seems as much related to factors in
the patient-doctor relationship or in the
personality of the individual mother as in
anything else. Third, studies which purport
to show advantages of one treatment
regi-men over another or seek to define optimal duration of therapy must include particular
care in evaluating how well the patients
have taken the medicine. Is the consistent superiority of intramuscular benzathine
penicillin over various oral therapy
regi-mens in the treatment of streptococcal pharyngitis due to the drug itself or the fact
that its take rate is assured? Is 10 days of
oral therapy truly superior to 7 days, or are patients more likely to take the medicine for at least 7 days if medicine is prescribed for a longer time? The evidence presented here does not presume to answer these
questions but merely indicates a need for
their clarification. Finally, we are once
again impressed with the fact that the phy-sician’s responsibility does not end with the
prescription of appropriate therapy and
must extend to a concern with the
practi-cality and acceptance of the therapy. In light of this, the alternative of a single in-tramuscular injection of benzathine penicil-un is to be seriously considered, especially if the physician does not know the family vell, or if the child does not appear very ill.
SUMMARY
A study of how well children took oral
penicillin when prescribed for streptococcal pharyngitis or otitis media was conducted in three private pediatric group practices.
The presence of penicillin in the urine, as
determined by the Sarcina lutea culture
method, was used to assess the compliance rate. A total of 459 patients were
studied-107 at five days of therapy and 352 at nine days.
Eighty-one percent of the patients were
taking the penicillin as prescribed on the
fifth day, and 56% were taking it on the
ninth day. Another 13% were erratic takers
at the end of therapy. The children with
pharyngitis were somewhat more likely to
complete therapy than those with otitis
EDIT0II’s NOTE: See the Commentary beginning on page 157 of this issue. an attempt to differentiate those who
com-plied with the therapy from those who did
not. The take rate was unrelated to age or
sex of the child, duration of symptoms, or doctor’s estimate of severity of disease at onset. It was correlated with mother’s
esti-mate of severity, whether or not their
usual doctor prescribed the medicine, and certain personality traits of the mother as perceived by the pediatrician. The implica-lions of the study for the use of oral thera-py in children are discussed.
REFERENCES
1. Davis, M. S., and Eichhorn, R. L. : Compliance
with medical regimen: A panel study. J. Health Hum. Behavior, 4:240, 1963.
2. Mohler, D. N., Wallin, D. C., and Dreyfus,
E. C.: Studies in home treatment of
strep-tococcal disease, I. New Eng. J. Med., 252:
1116, 1955.
3. Mohler, D. N., Wallin, D. C., Dreyfus, E. C.,
and Bakst, H. J.: Studies in home
treat-ment of streptococcal disease II. New Eng.
J. Me(l., 254:45, 1956.
4. Feinstein, A. B., \Vood, II. F., Epstein, J. P.,
Taranta, A., Simpson, R., and Tursky E.:
Controlled study of three methods of
pro-phylaxis against streptococcal infection in
pop-ulation of rheumatic children II. New Eng. J.
Med., 260:697, 1959.
5. Bergman, A. B., and Werner, R. J.: Failure of
children to receive penicillin by mouth. New Eng. J. Med., 268:1334, 1963.
6. Leistyna, j. A., and Macauley, J. C. : Therapy
of streptococcal infections. Amer. J. Dis.
Child., 111:22, 1966.
7. Jackson, H., Cooper, J., Nellinger, W., and
Olsen, A. : Streptococcal pharyngitis in rural
practice. J.A.M.A., 197:385, 1966.
8. Crove, D. C., and Randall, W. A. : Assay
I’Iethods of Antibiotics: A Laboratory
Man-ual. New York: Medical Encyclopedia, Inc.,
Chapter 2, 1958.
9. Cough, H. C. : The adjective check list as a
personality assessment research technique.
Psychol. Rep., 6: 107, 1960.
Acknowledgment
The authors wish to thank Dr. Lowell Glasgow
for his helpful suggestions regarding the S. lutea
technique and Dr. Robert Haggertv for his advice
and encouragement in establishing the
collabora-tive research program and his guidance with this