nature publishing group
Because of the well-known limitations of preapproval clinical
trials, the safety profile of a drug is only partly known at the
time of market entry.1 Market approval is a signal not of the end
of drug development but of the start of continuous evaluation
of both benefits and risks during the entire life cycle of a drug.
Throughout this life cycle, serious safety issues may emerge2–4
that can cause hospitalization, disability, or even death of
patients.5,6 Health-care professionals need to be informed of
these safety issues as soon as possible in order to minimize the
risk of preventable adverse drug events. In the European Union,
these risks are communicated through paper-based warning let-
ters, so-called Direct Health-Care Professional Communications
(DHPCs), or “Dear doctor” letters. Over the past decade, risk
minimization interventions such as DHPCs have been issued
in increasing numbers2,7 to ensure continued safe and effective
use of medicinal products.8
However, the limited evidence so far indicates that DHPCs
are not always effective in changing the behavior of physi-
cians.9,10 Most studies that have assessed the impact of drug
safety warnings focused on one drug, or only on a limited
number of warnings, and they often had methodological
limitations.9 When looking at a large number (58) of various
drug safety issues, we showed that DHPCs lowered drug use in
half of the cases in the short term and in a third of the cases in
the long term.11
With the new European Union pharmacovigilance legis-
lation that came into force in July 2012, evaluation of the
impact of risk minimization measures has become manda-
tory.12,13 Currently, it is unknown which determinants might
influence the impact of DHPCs. A better understanding of
the influence of these determinants can facilitate optimiza-
tion of future risk communication and evaluation of risk
minimization measures. In this study, we explore the impact
of drug- and DHPC-related characteristics on the effect of
DHPCs on drug use.
We identified 59 DHPCs for 46 drugs that fulfilled all inclusion
and exclusion criteria. Two DHPCs that were issued within two
consecutive months for nelfinavir were analyzed as one. This
resulted in 58 evaluable drug and DHPC pairs for which the rela-
tive changes in new drug use following the DHPC were calculated
Serious safety issues relating to drugs are communicated to health-care professionals via Direct Health-Care Professional
Communications (DHPCs). We explored which characteristics determined the impact of DHPCs issued in the Netherlands
for ambulatory-care drugs (2001–2008). With multiple linear regression, we examined the impact on the relative
change in new drug use post-DHPC of the following: time to DHPC, trend in use, degree of innovation, specialist drug,
first/repeated DHPC, DHPC template, and type of safety issue. DHPCs have less impact on use of specialist drugs than
nonspecialist drugs (P < 0.05). The DHPCs' impact increased after availability of a template emphasizing the main
problem (P < 0.05), and for safety issues with a risk of death and/or disability (both P < 0.05) (adjusted R² = 0.392). Risk
communication can be effective, specifically in case of well-structured information, and very serious safety issues.
Effectiveness may improve by tailoring DHPCs and adding other communication channels, for example for drugs that are
increasingly being used.
Received 21 August 2012; accepted 11 December 2012; advance online publication 27 February 2013. doi:10.1038/clpt.2012.262
When Direct Health-Care Professional
Communications Have an Impact on
Inappropriate and Unsafe Use of Medicines
KC Reber1, S Piening2, JE Wieringa1, SMJM Straus3,4, JM Raine5, PA de Graeff2,3, FM Haaijer-Ruskamp2
and PGM Mol2,3
The first two authors contributed equally to this work.
1Department of Marketing, University of Groningen, Groningen, The Netherlands; 2Department of Clinical Pharmacology, University of Groningen, University Medical
Center Groningen, Groningen, The Netherlands; 3Dutch Medicines Evaluation Board, Utrecht, The Netherlands; 4Department of Medical Informatics, Erasmus Medical
Center, Rotterdam, The Netherlands; 5Medicines and Healthcare Products Regulatory Agency, London, UK. Correspondence: PGM Mol (email@example.com)
CliNiCal PHaRmaCology & THERaPEuTiCS
(Table 1). The median number of new drug users per month in
the year before issuing the DHPC ranged from 7 (sirolimus) to
53,596 (salbutamol) (Supplementary Table S1 online).
The mean relative change in new drug use among all the
DHPCs analyzed was −9% (SD: ±24%) and ranged from −67.4%
for strontium ranelate to +71.7% for sirolimus (Supplementary
Table S1 online). The median time from approval to DHPC was
82.5 months (6.9 years, interquartile range: 3.4–13.6 years), and
80% of the DHPCs were issued for drugs that had been licensed
for more than 3 years. Almost a quarter of the drugs showed a
decrease in new drug use before the DHPC. Similar numbers of
DHPCs were issued for all drugs independent of their degree of
innovation (important, moderate, and solely pharmacological/
technological), with a few drugs classified as modestly innova-
tive. More than half (59%) of the DHPCs were sent for special-
ist drugs. The majority (71%) of the 58 DHPCs concerned a
first DHPC. DHPCs were evenly divided over the seriousness
When the first block with the drug characteristics was entered
in the model to test if these characteristics explained any dif-
ferences in the impact of the DHPC on drug use, we found that
DHPCs sent for specialist drugs were associated with a more
positive change in use than the change in use of nonspecialist
drugs (Table 2; model 1, P = 0.046). Within the group of drugs
for which the DHPC leads to a decrease in use, the positive β
value indicated that the negative usage effect was (partially)
offset for specialist drugs. Conversely, for the cases where a
DHPC increased drug use, the increase was stronger for spe-
cialist drugs than for nonspecialist drugs. This effect remained
significant after entering the DHPC characteristics into the
model (Table 2; model 2, P = 0.008). In the second model, we
also found that DHPCs for drugs with a decreasing pre-DHPC
trend were associated with a change toward lower drug use; this
effect was marginally significant (Table 2; model 2, P = 0.055).
DHPCs issued after a template was made available contributed
to a change toward lower drug use (Table 2; model 2, P < 0.05).
The safety issues related to risk of death as well as disability
were significantly associated with changes toward lower drug
use (Table 2; model 2, P < 0.05 for both), whereas no signifi-
cant impact was observed for safety issues regarding the risk of
hospitalization (Table 2; model 2, P = 0.867).
The block of DHPC characteristics contributed significantly to
the model, explaining an additional 32% of variance (F change =
5.906, ∆R² = 0.315, P ≤ 0.001). The drug and DHPC characteris-
tics together explained 39% (adjusted R² = 0.392) of the overall
variation in change of new drug use.
This study gives a first impression of the determinants that
increase the impact of DHPCs on drug use. We found that
declining drug use before the issuance of a DHPC, specialist
drugs, the type of serious safety issue, and DHPCs that were
issued after the DHPC template was made available were all
associated with changes in drug use. We discuss the comments
from the viewpoint of the most common situation, in which a
DHPC leads to a decrease in the number of new users.
The marginally significant effect found for already declining
use pre-DHPC confirms our earlier assumption that DHPCs
have an accelerating effect on the decline in use of drugs that
are at the end of their life cycle, when several substitute drugs
have become available.11
As previously hypothesized,11 we observed that DHPCs issued
for drugs that require a specialist to initiate prescribing have
less impact than those sent for drugs that can also be prescribed
by a general practitioner. Drugs are given this requirement in
the summary of product characteristics because of the expected
Table 1 Descriptive statistics for outcome and independent
Outcome measure (relative change in new drug use)
Time to DHPC since registration
Median, years (IQR)
≤3 years (%)
>3–10 years (%)
>10 years (%)
Trends in use (before DHPC was issued), n (%)
No change in use
Degree of therapeutic innovation, n (%)
Specialist drug, n (%)
First/repeated DHPC, n (%)
DHPC template, n (%)
Type of serious safety issue, n (%)
(Prolonged) hospital admission
(Temporary/persistent) disability or incapacity/
Drug and DHPC paira
−0.674 to 0.717
DHPC, direct health-care provider communication; IQR, interquartile range.
aUnless otherwise indicated, data are reported as numbers (percentages) of drug and
DHPC pairs. Percentages might not add up to 100% due to rounding.
complexity in prescribing them. The specialist drugs in our
sample were prescribed mainly for HIV infection, epilepsy, and
cancer. Specialists often have additional resources at their dis-
position to monitor their patients, which facilitates continued
use of a drug post-DHPC.14 Also, specialists might tend to use
more risky treatment options in view of the patient population
they treat (e.g., more complex patients, patients who previously
failed on other therapies). Another explanation could be that
the perception of their own expertise limits their willingness to
accept recommendations from others, as was observed during
implementation of treatment guidelines.15 Toward the end of
our study period, the European guidelines were amended to
include a fixed DHPC template.8 When we analyzed the content
of the DHPCs in our sample, we observed an increase in uni-
formity of the structure and layout of the DHPCs. The results
of our analysis confirm that DHPCs issued after the DHPC
template was made available had more impact than did DHPCs
issued before the availability of the template, suggesting that the
template has contributed to the understandability and uptake
of the safety information, which would be in line with earlier
findings that explicit wording contributes to improved uptake
of DHPC recommendations.16 The increased impact of more
recent DHPCs also reflects a generally increased risk aware-
ness due to intensified media accessibility and rapid information
availability.17 Extensive media attention for certain drug safety
issues (e.g., for rofecoxib and rosiglitazone) and more proactive
pharmacovigilance in the past decade may have further con-
tributed to the awareness of prescribers regarding drug safety
Communicating on serious safety issues potentially caus-
ing death or disability led to significantly lower drug use. Even
though all DHPCs are issued for serious safety issues, it is to be
expected that these particularly serious safety issues will have a
greater effect on the prescribing behavior of physicians.11
The impact of DHPCs is not influenced by the age of the
drug, indicating that DHPCs affect the use of older and newer
drugs in the same way. This is consistent with our earlier find-
ing that important safety issues requiring DHPCs are identified
Table 2 Results of the weighted regression analysis
Blocks entered in the analysis
model 1 (block 1 entered)model 2 (both blocks entered)
B (95% Ci)β
P value B (95% Ci)β
(Constant)−0.214 (−0.386; −0.042)
3.05 × 10−4 (−0.000; 0.001)
0.016 −0.043 (−0.212; 0.126)
1.67 × 10−4 (−0.000; 0.001)
Time to DHPC (months)0.149 0.257 0.0820.478
Trend in use (before DHPC)
Increasing Ref RefRefRef
No change0.036 (−0.097; 0.169)0.0840.593 0.011 (−0.106; 0.128)0.0250.854
Decreasing −0.080 (−0.242; 0.082)−0.1600.324 −0.135 (−0.273; 0.003)−0.2700.055
Degree of therapeutic
−0.002 (−0.057; 0.052)−0.0130.9340.004 (−0.044; 0.053)0.0250.860
NoRef Ref Ref Ref
Yes 0.136 (0.002; 0.269)0.3200.046 0.159 (0.043; 0.274)0.373 0.008
Repeated −0.092 (−0.213; 0.029)−0.1870.133
Yes −0.157 (−0.266; −0.049) −0.3080.005
Type of serious safety issue
Death −0.265 (−0.415; −0.114)−0.4500.001
Hospital admission−0.012 (−0.152; 0.129)−0.025 0.867
−0.155 (−0.288; −0.023)−0.3480.023
R² (adjusted R²)0.184 (0.106)0.499 (0.392)
F change 5.9060.000
A negative regression coefficient is associated with a lower use post-DHPC (i.e., a larger decrease or a smaller increase as a result of the DHPC).
B, raw regression coefficient; β, standardized regression coefficient; CI, confidence interval; DHPC, Direct Health-care Professional Communication; Ref, reference category.
aTreated as continuous predictor.
CliNiCal PHaRmaCology & THERaPEuTiCS
throughout the entire life cycle of drugs,2 which would indicate
that the age of the drug does not need to be considered when
tailoring the communicating drug safety issues.
More innovative drugs did not show greater impact of a
DHPC on drug use than less innovative drugs. Therapeutically,
innovative drugs can provide physicians with treatment options
for complex patients who do not respond well to less innovative
drugs. Physicians could be of the opinion that the innovativeness
of the drug outweighs the risk of occurrence of the safety issue.
However, our level of analysis does not allow us to elaborate how
this translates to behavior of individual physicians. This aspect
could be explored in a focus-group setting or by conducting
individual interviews with prescribers.
Our results show that a repeated safety warning is not necessarily
more effective in changing drug use than a single DHPC. This is
consistent with findings of several prior studies that reported no
changes in the assessed outcome after repeated safety warnings
were issued.19–22 The repeated DHPCs in our sample concerned
both identical and different safety issues. It is possible that repeated
DHPCs that were issued for the same safety issue are more effective
than repeated DHPCs that were issued for different safety issues
with the same drug. However, owing to the limited sample size, we
were not able to incorporate this aspect into our model.
Strengths and limitations
To our knowledge, this study is the first to systematically evalu-
ate determinants of the impact of DHPCs on new drug use. We
included a large number of DHPCs in our analyses, covering a
wide variety of drugs and safety issues. With the results of this
study, it will be possible to anticipate, and possibly enhance,
the impact of future DHPCs on drug use by tailoring risk com-
munication about safety issues of drugs more specifically. On a
case by case basis, it can be decided to add other communica-
tion channels in addition to the DHPC to obtain the desired
outcome: minimization of the risk. Communication tools addi-
tional to the DHPC should be considered for specialist drugs,
in case of safety issues leading to hospitalization, and for drugs
of which the use is on the rise or at a stable plateau before
the DHPC is issued. For example, an additional email could
prove to be a useful and easy tool to rapidly inform health-
care professionals of safety issues. Professional associations
and specialist learned societies should be involved in drafting
DHPCs for specialist drugs, for a more motivating outreach to
reduce prescribing if that is the desired outcome. Repetition of
the message for example by professional associations to their
members, either by email or in their news bulletins could also
improve the impact of the communication. Although in case
of the most severe safety issues that can lead to disability and/
or death of the patient, one clear and strong DHPC might be
sufficient. To make the DHPC stand out more to health-care
professionals, they should all be sent with an extra symbol
such as a picture of an orange hand printed on the envelope,
as is currently done in the Netherlands in cases that require
immediate action from the health-care professional. A dif-
ferent color, e.g., yellow could be used to distinguish DHPCs
to merely alert HCPs of a drug safety issue from cases where
immediate action is required. We included a set of seven fac-
tors in our full model; however, the range of determinants is
limited by the sample size and its corresponding power. Our
full model explained 39% of the overall variation in DHPC
effect size and can be considered a first exploration of determi-
nants that influence the impact of DHPCs. Other factors that
we did not include in our model might also cause variations
in the impact of DHPCs, for example, media attention, the
incidence of safety issues, safety issues related to off-label use,
and availability of an alternative treatment. It has been sug-
gested that media attention can play an important role in the
impact of DHPCs.23 To probe this, we performed an explora-
tory lay- and professional-literature search for a selection of
the DHPCs in our study population. This search resulted in
too little information to include presence of media attention
in our model. Likewise, the incidence of the safety issue could
not be included because this aspect was not mentioned in the
majority of the DHPCs. Too few DHPCs concerned safety
issues related to off-label use, which led to insufficient vari-
ation within the variable for incorporation into our model.
Alternative treatment was available for almost all drugs and
was indirectly covered in the innovation variable. We did not
find associations for older vs. newer drugs and degree of inno-
vation with DHPC impact. Therefore, it seems unlikely that
availability of an alternative treatment is a major determinant.
The limited sample size could be addressed by repeating this
study in a few years, when more DHPCs will have been issued.
In addition, our study was limited to the Dutch setting, and
extrapolation of these findings to hospital drugs is not possible. A
European Union–wide study would allow for a comparison of the
impact of DHPCs as well as the determinants of impact of DHPCs
in different countries. This will provide much needed information
regarding locally tailored risk-communication strategies.
It should be noted that a decrease in use is not always the
desired impact of a DHPC. The results of this study can there-
fore be used only to anticipate the impact of DHPCs on new
drug use, not for other outcomes that might be more attuned
to the recommendation in the DHPC, such as necessity for
liver-function tests performed in case of hepatotoxicity risk.
This means that any additional action should be carefully con-
sidered. Nevertheless, we think that new drug use is the most
appropriate outcome measure with which to explore the role
of determinants of impact of DHPCs because it is the one out-
come measure that can reliably be assessed for such a large
group of drugs. In addition, new use is a more sensitive meas-
ure than overall use because changes in prescribing behav-
ior are more likely with new users. Further research could be
aimed at clusters of drugs with the same recommendation in
the DHPC, e.g., all drugs that require laboratory testing or all
drugs with restrictions regarding concomitant use of contrain-
dicated drugs. This may provide insight into how the impact
of DHPCs on more specific outcomes can best be anticipated.
Conclusion and recommendation
This study provides a first exploration of determinants that influ-
ence the impact of DHPCs on drug use. The results show that
declining use before the DHPC, specialist drugs, DHPCs issued
after availability of a template, and type of serious safety issue
are associated with changes in new drug use. These results can
be used as a first step in more specifically tailoring risk com-
munication about safety issues of drugs.
Data collection. Data were collected for all drugs for which a DHPC
was issued in the Netherlands between January 2001 and January 2008.
Monthly dispensing data for the period 2000–2008 were obtained from
the Dutch Foundation for Pharmaceutical Statistics. Its database contains
drug dispensing data of >95% of Dutch community pharmacies, serv-
ing ~15.3 million people.24 The DHPCs were collected from the website
of the Dutch Medicines Evaluation Board25 and the Dutch Healthcare
Inspectorate paper archive. We excluded DHPCs for drugs that were
not dispensed in ambulatory care, drugs with insufficient dispensing
data (no more than 10 prescriptions/month for new users, who were
not prescribed the same drug within the previous 6 months, pre- and
post-DHPC), and drugs for which a market withdrawal was announced
in the DHPC.
The drug and DHPC characteristics were retrieved from the DHPCs,
the human medicines database of the Dutch Medicines Evaluation
Board,26 the World Health Organization’s Anatomical Therapeutic
Chemical classification system, and version 13 of the Medical Dic-
tionary for Regulatory Activities (MedDRA; the international medical
terminology developed under the auspices of the International Con-
ference on Harmonisation of Technical Requirements for Registration
of Pharmaceuticals for Human Use). We recorded the international
nonproprietary names, Anatomical Therapeutic Chemical classifica-
tion, registration date, date of DHPC, and safety issue (including system
Outcome measure. The outcome measure for this study was the relative
change in new drug use after a DHPC was issued. We defined new drug
use as the number of new prescriptions of a drug for which no prescrip-
tions had been dispensed to the patient in the previous 6 months. We
chose new drug use as our outcome measure because we assumed it
to be more sensitive to changes in prescribing than overall drug use.
The relative change was calculated as the absolute change in drug use
divided by the median drug use in the 12 months before the DHPC.
Changes in the absolute number of new drug use were determined
through interrupted time series analyses based on separate autore-
gressive integrated moving average models for each individual drug.
Observed changes indicated a change in the level of new use from the
time of the DHPC to the end of the observation period. The calculation
of the outcome measure and inclusion and exclusion criteria have been
described in more detail elsewhere.11
Determinants. Characteristics of the drugs and the DHPC were assessed
to explain differences in the outcome. We included four drug-related
1. The time to DHPC, defined as the elapsed time in months from
drug approval (registration date) to the publication of the DHPC.
2. Trends in use before the DHPC, based on trend analyses11 to
identify changes in the number of new users in the 12 months
before the publication of the DHPC.
3. The degree of therapeutic innovation, determined by using the
score of therapeutic innovation as reported by Motola et al. for
drugs that were centrally approved in Europe.27 Using this score,
drugs can be classified as important, moderate, modest, or as solely
pharmacological/technological innovations, taking into account
the seriousness of the disease, the availability of alternative drugs,
and whether drug effects have been shown on relevant clinical end
points and observed effect size. For the drugs that were approved
via the decentralized system, that is, at the national level, two
investigators (P.G.M.M. and P.A.d.G.) independently evaluated the
degree of therapeutic innovation using the “Motola algorithm.” In
case of disagreement, consensus was reached by discussion.
4. Specialist drugs; i.e., the drug required an initial prescription
from a medical specialist as indicated in the summary of product
The following three DHPC-related characteristics were included:
1. First or repeated DHPC, a dichotomous variable indicating whether
the DHPC was the first safety-related regulatory action or whether
another DHPC had been sent previously—this included identical as
well as different safety issues;
2. DHPC template, a dichotomous variable indicating whether the
DHPC was issued after a DHPC template was published in Volume
9A of “The Rules Governing Medicinal Products in the European
Union” in January 2007; and
3. The type of serious safety issue, which was classified according to
the World Health Organization list of serious adverse events or
reactions, as resulting in death, (prolongation of) hospitalization, or
persistent or significant disability/incapacity.28
We added an “other” category for cases that could not be classified into
any of the aforementioned categories, e.g., product contamination. Two
pharmacovigilance experts (medical doctors) independently categorized
the adverse drug reactions. Any disagreement was resolved by a third
Statistical analyses. We performed a multiple linear regression analysis to
examine the impact of drug and DHPC characteristics on the observed
relative change in new drug use following a DHPC. Because the assump-
tion of homoscedasticity, one of the key assumptions in linear regres-
sion,29 was not fulfilled, a weighted least-squares procedure was applied.
The weight is inversely related to the uncertainty of the information con-
tained in the associated data point. The point estimates of relative changes
in new drug use weighed less when the observed absolute changes in
effect sizes were found to have higher SEs in the autoregressive integrated
moving average model. The independent variables were entered block-
wise, with the variables describing the drug characteristics entered in the
first block. The second block included the DHPC characteristics. The
degree of therapeutic innovation was treated as a continuous, independ-
ent variable in the analysis. The explained variance of the model was indi-
cated by the adjusted R². The significance of each block was tested using
F change, and the contribution of each block to the variance explained
was computed (∆R²). Raw coefficients (B) with 95% confidence intervals,
standardized β coefficients (β), and P values were calculated. Statistical
analyses were performed with SPSS 18 (SPSS, Chicago, IL).
SUPPLEMENTARY MATERIAL is linked to the online version of the paper at
This study was supported by an unconditional grant from the Dutch
Medicines Evaluation Board.
K.C.R., S.P., J.E.W., S.M.J.M.S., J.M.R., P.A.d.G., F.M.H.-R., and P.G.M.M. wrote
the manuscript; K.C.R., S.P., J.E.W., S.M.J.M.S., P.A.d.G., F.M.H.-R., and P.G.M.M.
designed the research; K.C.R., S.P., J.E.W., P.A.d.G., and P.G.M.M. performed
the research; and K.C.R., S.P., J.E.W., S.M.J.M.S., J.M.R., P.A.d.G., F.M.H.-R., and
P.G.M.M. analyzed the data.
CONFLICT OF INTEREST
S.M.J.M.S., P.A.de G., P.G.M.M., and J.M.R. are also employees of the Dutch
Medicines Evaluation Board and UK MHRA. The views presented in this
article are their personal views and do not necessarily reflect those of the
regulatory agencies or one of their committees or working parties. The other
authors declared no conflict of interest.
CliNiCal PHaRmaCology & THERaPEuTiCS
Articles Download full-text
© 2013 American Society for Clinical Pharmacology and Therapeutics
1. Stricker, B.H. & Psaty, B.M. Detection, verification, and quantification of
adverse drug reactions. BMJ 329, 44–47 (2004).
Mol, P.G., Straus, S.M., Piening, S., De Vries, J.T., De Graeff, P.A. &
Haaijer-Ruskamp, F.M. A decade of safety-related regulatory action in the
Netherlands: A retrospective analysis of direct healthcare professional
communications from 1999 to 2009. Drug Saf. 33, 463–474 (2010).
Giezen, T.J., Mantel-Teeuwisse, A.K., Straus, S.M., Schellekens, H., Leufkens, H.G.
& Egberts, A.C. Safety-related regulatory actions for biologicals approved in
the United States and the European Union. JAMA 300, 1887–1896 (2008).
Lasser, K.E., Allen, P.D., Woolhandler, S.J., Himmelstein, D.U., Wolfe, S.M. & Bor,
D.H. Timing of new black box warnings and withdrawals for prescription
medications. JAMA 287, 2215–2220 (2002).
Pirmohamed, M. et al. Adverse drug reactions as cause of admission to
hospital: prospective analysis of 18 820 patients. BMJ 329, 15–19 (2004).
Sari, A.B. et al. Extent, nature and consequences of adverse events: results of
a retrospective casenote review in a large NHS hospital. Qual. Saf. Health Care
16, 434–439 (2007).
Nkeng, L., Cloutier, A.M., Craig, C., Lelorier, J. & Moride, Y. Impact of regulatory
guidances and drug regulation on risk minimization interventions in drug
safety: a systematic review. Drug Saf. 35, 535–546 (2012).
European Commission. Volume 9A of the rules governing medicinal products
in the European Union. Guidelines on Pharmacovigilance for medicinal
products for human use <http://ec.europa.eu/health/files/eudralex/vol-9/
pdf/vol9a_09-2008_en.pdf> (2008). Accessed 13 July 2012.
Piening, S. et al. Impact of safety-related regulatory action on clinical practice:
a systematic review. Drug Saf. 35, 373–385 (2012).
10. Dusetzina, S.B. et al. Impact of FDA drug risk communications on health care
utilization and health behaviors: a systematic review. Med. Care 50, 466–478
11. Piening, S. et al. Impact of safety-related regulatory action on drug use in
ambulatory care in the Netherlands. Clin. Pharmacol. Ther. 91, 838–845 (2012).
12. The European Parliament and the European Council. Regulation (EU) no.
0:348:0001:0016:EN:PDF> (2010). Accessed 13 July 2012.
13. The European Parliament and the European Council. Directive 2010/84/EU
0099:EN:PDF> (2010). Accessed 13 July 2012.
14. Harrold, L.R., Field, T.S. & Gurwitz, J.H. Knowledge, patterns of care, and
outcomes of care for generalists and specialists. J. Gen. Intern. Med. 14,
15. Kasje, W.N., Denig, P. & Haaijer-Ruskamp, F.M. Specialists’ expectations
regarding joint treatment guidelines for primary and secondary care. Int. J.
Qual. Health Care 14, 509–518 (2002).
16. Weatherby, L.B., Nordstrom, B.L., Fife, D. & Walker, A.M. The impact of wording
in “Dear doctor” letters and in black box labels. Clin. Pharmacol. Ther. 72,
17. Ropeik, D. & Slovic, P. Risk communication: a neglected tool in protecting
public health. Risk in Perspective 11, (2003).
18. Raine, J., Wise, L., Blackburn, S., Eichler, H.G. & Breckenridge, A. European
perspective on risk management and drug safety. Clin. Pharmacol. Ther. 89,
19. Kurdyak, P.A., Juurlink, D.N. & Mamdani, M.M. The effect of antidepressant
warnings on prescribing trends in Ontario, Canada. Am. J. Public Health 97,
20. Kurian, B.T., Ray, W.A., Arbogast, P.G., Fuchs, D.C., Dudley, J.A. & Cooper, W.O.
Effect of regulatory warnings on antidepressant prescribing for children and
adolescents. Arch. Pediatr. Adolesc. Med. 161, 690–696 (2007).
21. Olfson, M., Marcus, S.C. & Druss, B.G. Effects of Food and Drug Administration
warnings on antidepressant use in a national sample. Arch. Gen. Psychiatry 65,
22. Jones, J.K., Fife, D., Curkendall, S., Goehring, E. Jr, Guo, J.J. & Shannon, M.
Coprescribing and codispensing of cisapride and contraindicated drugs.
JAMA 286, 1607–1609 (2001).
23. Martin, R.M., May, M. & Gunnell, D. Did intense adverse media publicity impact
on prescribing of paroxetine and the notification of suspected adverse drug
reactions? Analysis of routine databases, 2001–2004. Br. J. Clin. Pharmacol. 61,
24. Foundation for Pharmaceutical Statistics <http://www.sfk.nl/english>.
Accessed 13 July 2012.
25. Dutch Medicines Evaluation Board. Direct Healthcare Professional
pharmacovigilance/DHPC/default.htm> (2012). Accessed 13 July 2012.
26. CBG-MEB. Database human medicines <http://www.cbg-meb.nl/CBG/en/
Accessed 13 July 2012.
27. Motola, D., De Ponti, F., Rossi, P., Martini, N. & Montanaro, N. Therapeutic
innovation in the European Union: analysis of the drugs approved by
the EMEA between 1995 and 2003. Br. J. Clin. Pharmacol. 59, 475–478
28. World Health Organization Collaborating Centre for International Drug
Monitoring. Glossary of terms in pharmacovigilance <http://www.who-umc.
org/graphics/25301.pdf> (2011). Accessed 13 July 2012.
29. Greene, W.H. Econometric Analysis 6th edn. (Prentice Hall, Upper Saddle River,
WHAT IS THE CURRENT KNOWLEDGE ON THE TOPIC?
3 DHPCs are not always effective in changing the clini-
cal behavior of physicians. Approximately half of the
DHPCs lower drug use in the short term and in a third
of cases in the long term.
WHAT QUESTION DID THIS STUDY ADDRESS?
3 We explored which characteristics determine impact of
WHAT THIS STUDY ADDS TO OUR KNOWLEDGE
3 A declining use prior to the DHPC, specialist drugs,
DHPCs issued after the availability of a DHPC template,
and the type of serious safety issue are associated with
changes in new drug use.
HOW THIS MIGHT CHANGE CLINICAL PHARMACOLOGY
3 Our overview will help to develop more tailored
approaches to communicating drug safety issues to