Professional Documents
Culture Documents
ORIGINAL ARTICLE
improved preventive care, reduced hospitalization Of all Social Health Insurance Act beneficiaries,
and lower costs (2, 4–9). 338 423 (79Æ6%) had at least one pharmacy claim
Continuity of care has been addressed espe- during 2001. Of these, all patients visiting (This
cially from the perspective of general medical means that a visit was made to a pharmacy con-
practice (10). Little is known about the relation- cluded by a dispensing and a pharmacy claim to
ship between the continuity of care from a the health insurance company based upon the
community pharmacy perspective and clinical dispensing.) two or more pharmacies in 2001 –
outcomes. Some studies have described pharma- thus having received at least two prescriptions
cists’ provision of continuity of care for special in 2001 – were identified from the pharmacy
groups of patients (e.g. HIV patients) or the claims database (n = 45 805). These patients were
provision of structures that support continuity of indicated as ‘shoppers’. From all the other bene-
care across health care settings (e.g. transfer of ficiaries who received at least one prescription
information about drug use) (11–14). Discontinu- and visited only one pharmacy during 2001
ity of pharmacy care may put the patient at risk (n = 292618), a random, numerically equivalent,
for drug therapy-related problems, because sample was taken (n = 45 805). These patients
pharmacy shopping hampers adequate medica- were indicated as ‘non-shoppers’.
tion surveillance. Examples of such unintentional For shoppers as well as non-shoppers, data were
problems not only include unwanted duplicate obtained comprising age and gender. For each
medications, drug–disease interactions, drug patient, a medication history was collected cover-
intolerabilities and drug–drug interactions, but ing information about all dispensed and (partially)
also conflicting information about drug use from reimbursed drugs during 2001, such as name,
different pharmacies, confusion between brand Anatomical Therapeutic Chemical (ATC) code,
and generic names and incorrect quantities (3, date of dispensing, dispensed amount, dosage
15). regimen, type of prescriber and the community
As continuity of pharmacy care can be consid- pharmacy (anonymous, unique code) where the
ered an important prerequisite for the clinical risk drug had been dispensed.
management of drug therapy-related problems, we
studied the prevalence and determinants of phar-
Classification of data
macy shopping behaviour.
Shoppers were classified into three mutually
exclusive categories based upon (i) the number of
METHODS visits to one or more pharmacies other than the
main dispensing pharmacy (=‘elsewhere’), (ii) the
Setting, study population and data collection
proportion of prescriptions dispensed in pharma-
Data were obtained from the pharmacy claims cies elsewhere and (iii) the total number of pre-
database of a Dutch health insurance company scriptions dispensed in pharmacies elsewhere
(‘O.W.M. Zorgverzekeraar Zorg en Zekerheid u.a.’) (Table 1)1. Several characteristics were investigated
concerning the year 2001. This health insurance as determinants of shopping behaviour: gender,
company mainly proceeds in the western region age (four categories: 0–25; 26–40; 41–60; >60), the
between The Hague and Amsterdam, and can be number of different type of prescribers [general
described as relatively small (on average 448 392 practitioner (GP), specialist or other], and the
beneficiaries in 2001). number of different drugs (active substances) dis-
The data obtained were related to the beneficia- pensed in 2001 (based upon ATC code-level 7;
ries who were insured under the Social Health three categories: 0–2; 3–5; >5). In addition, we
Insurance Act comprising all employees earning studied whether any dispensing in 2001 of the
less than about 33 000 Euro per year, social security therapeutic groups of drugs in accordance with the
recipients and certain old-age groups. In 2001, ATC classification of the WHO Collaborating
about 65% of the Dutch population was insured 1
The main pharmacy means the pharmacy in which more
under this law, and 94Æ8% (on average 425 061) prescriptions were dispensed for the concerning patient than
within this insurance company. in each other pharmacy separately.
2008 The Authors. Journal compilation 2008 Blackwell Publishing Ltd, Journal of Clinical Pharmacy and Therapeutics, 33, 17–23
Prevalence and determinants of pharmacy shopping behaviour 19
a
‘Visited’ means that a visit was made to a pharmacy concluded by a dispensing and a pharmacy claim to the health insurance company
based upon the dispensing (see the Methods section).
Centre for Drug Statistics Methodology (level 1), pharmacy only once (data not shown). A small
as well as subclasses of the Nervous System drugs minority (0Æ2%) of the shoppers was classified as
(N category), was associated with shopping ‘heavy shoppers’.
behaviour. Table 2 shows the characteristics of the study
population. Comparing all shoppers with non-
shoppers and adjusted for all included variables,
Data analysis
female gender (ORadj 1Æ2; 95%CI 1Æ1–1Æ2), younger
Data were analysed using standard descriptive age (£40 year) (ORadj 1Æ7; 95%CI 1Æ7–1Æ8), the use of
data analysis (SPSS version 12Æ0). Logistic regression three or more different drugs (ORadj 2Æ9; 95%CI
analysis was used to estimate the strength of the 2Æ8–3Æ0) and different kind of prescribers (ORadj 2Æ4;
association between characteristics and pharmacy 95%CI 2Æ4–2Æ5) were associated with shopping
shopping behaviour and expressed as odds ratios behaviour.
(OR) with 95% confidence intervals (CI). This work Shoppers received more frequently at least one
was conducted in compliance with the require- prescription for systemic anti-infectives (51Æ7% vs.
ments of the Institutional Review Board of the 30Æ8%; OR 2Æ4; 95%CI 2Æ3–2Æ5) and for nervous
Department of Pharmacoepidemiology and Phar- system drugs (46Æ2% vs. 29Æ3%; OR 2Æ1; 95%CI 2Æ0–
macotherapy, Utrecht University. 2Æ1) than non-shoppers (Table 3). For the other
ATC classes, the differences were less clear. We
especially found a strong association between any
RESULTS dispensing of nervous system drugs and heavy
shopping (OR 16Æ7; 95%CI 9Æ1–30Æ5) as well as
Of the Social Health Insurance Act beneficiaries of
between any dispensing of nervous system drugs
the health insurance company (on average
and moderate shopping (OR 20Æ1; 95%CI 14Æ9–
n = 425 061 in 2001), a total number of 45 805
27Æ1). To some extent, similar associations were also
patients (10Æ8%) were identified who had visited
found for selected psychotropics, i.e. hypnotics and
more than one pharmacy in 2001 on at least one
anxiolytics, antidepressants, antipsychotics and
occasion (Table 1). Of these patients, the vast
opioids.
majority (98Æ8%) could be described as ‘light
shoppers’. Most of these ‘shopping’ patients
(86Æ4%) visited only one other pharmacy, 11Æ2% DISCUSSION
visited two and 2Æ4% three or more other phar-
10Æ8% beneficiaries were identified as pharmacy
macies. Within the group of patients visiting only
shoppers; 98Æ8% ‘light shoppers’, 1Æ0% ‘moderate
one other pharmacy, 63Æ4% visited the second
2008 The Authors. Journal compilation 2008 Blackwell Publishing Ltd, Journal of Clinical Pharmacy and Therapeutics, 33, 17–23
20 H. Buurma et al.
a
Active substances.
shoppers’ and 0Æ2% ‘heavy shoppers’. Pharmacy explanation is our finding that there was a strong
shopping occurred especially in women, younger association between any dispensing of systemic
people, people using a high number of different antibiotics and antimycotics and shopping behav-
drugs and those having different kind of prescrib- iour. These drugs are often needed in more or less
ers. The dispensing of any anti-infective drug was acute situations occurring during evenings, nights
related to (light) shopping. A strong association and in the weekend.
was found between any dispensing of nervous Nevertheless, also light shopping behaviour may
system drugs and heavy or moderate shopping. hamper adequate medication surveillance and put
This study confirms that Dutch patients are in the patient at risk for unintentional drug therapy-
general loyal to one pharmacy, leading to rather related problems, such as duplicate medications,
complete patient medication records (16). This may drug–disease interactions and drug–drug interac-
be due to the fact that in the Dutch health care tions. Not surprisingly, it has already been repor-
system, patients are historically closely linked to ted that an increasing number of pharmacists
one pharmacist. This situation is different from involved in the dispensing of drugs, increases the
several other countries. In the Netherlands, it is risk of dispensing potentially inappropriate drug
allowed to move around to seek medical treatment, combinations (15). In this respect, we may consider
especially outside office hours. Most shoppers the strong association between any dispensing of
visited only one other pharmacy. This ‘light’ systemic antibiotics and antimycotics and shop-
shopping behaviour is probably at least partly ping behaviour as an indication for a possible high
related to required pharmaceutical (and medical) frequency of unintentional, but potentially harmful
treatment outside office hours. Indicative for this drug–drug interactions, in which antibiotics,
2008 The Authors. Journal compilation 2008 Blackwell Publishing Ltd, Journal of Clinical Pharmacy and Therapeutics, 33, 17–23
Prevalence and determinants of pharmacy shopping behaviour 21
Table 3. Association between any dispensing of groups of medicines and shopping behaviour
particularly macrolides and fluoroquinolones, and Tamblyn et al. (15) found that the use of a single
several oral antimycotics, are involved (17). This dispensing pharmacy lowered the risk of potentially
warrants further investigation in future studies. inappropriate drug combinations.
The association between any dispensing of nervous To detect the problems due to pharmacy shop-
system drugs and shopping behaviour, especially ping, systems are needed which exchange infor-
heavy shopping behaviour (OR 16Æ7; 95% CI 9Æ1– mation among pharmacists. In the Netherlands,
30Æ5) and moderate shopping behaviour (OR 20Æ1; there is a tendency of locally and regionally clus-
95% CI 14Æ9–27Æ1) warrants new research as well, tering of pharmacy computer systems. The devel-
which we intend to do. The question is whether opment of a nationwide system, coordinated by the
there is a relation between shopping behaviour and Ministry of Health, is not expected to be finished
the heavy use of psychotropics. within the next 2–3 years. In a recent Canadian
The first step to reduce discontinuity of care due study, primary care physicians believed that such
to pharmacy shopping (which is frequently invisible an integrated system would improve continuity of
in the pharmacy) is better detection. Asking the care (18).
patient for actual medication use and diseases may This study had several limitations. In studies like
help to detect unintentional drug therapy-related ours, dispensing claims are considered to be iden-
problems, such as duplicate medications, drug– tical with the use of medicines. It is known, how-
disease interactions and drug–drug interactions, for ever, that not all dispensed drugs are used.
instance, those involving systemic antibiotics Secondly, pharmacy shopping might have been
and antimycotics. In addition, patients should be underestimated. A prescription, in some instances,
encouraged to stick not only to a single primary care may not have been followed by a dispensing,
physician, but also to a single dispensing pharmacy. because it was refused for some reasons, for
2008 The Authors. Journal compilation 2008 Blackwell Publishing Ltd, Journal of Clinical Pharmacy and Therapeutics, 33, 17–23
22 H. Buurma et al.
2008 The Authors. Journal compilation 2008 Blackwell Publishing Ltd, Journal of Clinical Pharmacy and Therapeutics, 33, 17–23
Prevalence and determinants of pharmacy shopping behaviour 23
2008 The Authors. Journal compilation 2008 Blackwell Publishing Ltd, Journal of Clinical Pharmacy and Therapeutics, 33, 17–23