The Association between Medication for Attention-Deficit/
Hyperactivity Disorder and Cancer
Hans-Christoph Steinhausen, MD, PhD, DMSc,1,2,3and Dorte Helenius, MSc1
Objective: A study on chromosomal abnormalities has raised concerns that medication with methylphenidate (MPH) for
attention-deficit/hyperactivity disorders (ADHD) might increase the risk of cancer. We performed a rigorous test of the
association between cancer and MPH and other drugs used for ADHD, based on data from nationwide Danish registers.
Methods: Data were linked from five registers containing information on a total of 21,186 patients with ADHD, their drug
a control group of patients with ADHD who had never taken medication. Frequencies of cancer diagnoses in these groups
were compared. In addition, hazard risk (HR) ratios for developing cancer, and survival rates in these subgroups, were
Results: The mean observation time varied between 1.3 and 10.8 years for the various drugs. Cancer rates in the total group
amountedto1.27per10 000person-yearsbeforeandto4.33per10000person-years afteronset oftreatment.Thefrequency
of cancer was significantly higher (p=0.05) after than before medication only in the antipsychotics subgroup. Furthermore,
for mixed medication, the cancer frequency in a small subgroup was significantly higher (p<0.05) after onset of medication
than in the unmedicated subgroup. The Cox regression analysis indicated that none of the drugs represented risk factors,
whereas higher dose (p<0.001) and older age (p<0.05) were risk factors for developing cancer.
developing cancer is not substantiated by these findings in a large and representative sample, which had been diagnosed and
treated over a period of 17 years.
of the most frequent mental disorders with a worldwide av-
erage prevalence of 5.3% in children (Polanczyk et al. 2007) and
3.4–4.4% in adults, according to recent United States and inter-
national surveys (Kessler et al. 2006; Fayyad et al. 2007). All in-
ternational guidelines agree that treatment with stimulants and
methylphenidate (MPH), in particular, play a major role in the
long-term professional care of the affected patients (Banaschewski
et al. 2006; Pliszka 2007). Therefore, prescription rates for cen-
trally acting drugs are high in many developed countries, with a
tendency to increase over the recent past (Zito et al. 2008; Zoega
et al. 2011).
Recently, concern has been raised by a study of chromosomal
abnormalities in peripheral lymphocytes in a small group of 12
children taking daily doses of MPH of between 20 and 54mg, with
each child serving as his or her own control and no use of a
ttention-deficit/hyperactivity disorder (ADHD) is one
concluded that three chromosomal abnormalities, namely, chro-
mosome aberration, sister chromatid exchanges, and micronuclei
remained unclear whether this well-documented biomarker for
cancer risk would actually translate into an increased risk of future
cancer in children receiving MPH treatment over an extended time
period. Furthermore, the study by El-Zein et al. (2005) has been
criticized for its serious experimental shortcomings, and both dif-
ficulties of interpreting its results and the uncertainties of its im-
plications for public health have been emphasized (Preston et al.
2005; Jacobson-Kram et al. 2008).
Following the same strategy, a first replication study in 12
to MPH studied the cytogenic effects on peripheral lymphocytes
before and 3 months after initiating MPH treatment (Ponsa et al.
2009). This study revealed no evidence of chromosomal aberra-
tions in their sample. Further studies found no increased frequency
months of treatment (Walitza et al. 2007), in an enlarged sample of
1Research Division for Child and Adolescent Psychiatry, Aalborg Psychiatric Hospital, Aarhus University Hospital, Aarhus, Denmark.
2Clinical Psychology and Epidemiology, Institute of Psychology, University of Basel, Basel, Switzerland.
3Department of Child and Adolescent Psychiatry, University of Zurich, Zurich, Switzerland.
JOURNAL OF CHILD AND ADOLESCENT PSYCHOPHARMACOLOGY
Volume 23, Number 3, 2013
ª Mary Ann Liebert, Inc.
31 children after 3 months of treatment (Walitza et al. 2009), and
after an extended observation period of 12 months in 12 children
(Walitza et al.2010).Noneofthe threestudiesrevealed an elevated
cytogenic damage. Another experimental study also failed to rep-
licate the genotoxic findings among 47 children after 3 months of
methylphenidate and amphetamine treatment (Witt et al. 2008).
The findings of all these experimental studies are limited be-
cause of relatively small sample sizes and short observation peri-
ods, so that the ultimate potential risk for developing cancer cannot
to elucidate the association of any cancer with MPH treatment in
only two epidemiological studies addressing this issue. An older
study identified 529 patients treated with MPH within a larger
follow up over 13–15 years, the study found no increased incidence
of cancer in this subgroup of MPH users (Selby et al. 1989).
A more recent study used a much larger cohort of patients in the
San Francisco bay and parts of California‘s Central Valley (Oes-
treicher et al. 2007). The authors compared cancer rates at 18 sites
among 35,400 MPH users who had taken it before age 20, to rates
among the local United States health provider membership in un-
exposed individuals (age, sex, and calendar year standardized), and
found 23 cancers among MPH users, versus 20.4 expected. MPH
use was associated with an increased risk of lymphocytic leukemia,
based on eight observed cases. The medical records of these ex-
posed cases did not reveal any lymphocytic leukemia risk factors
(prior cancer, radiotherapy or chemotherapy, or Down syndrome).
The authors concluded that there is no moderate or strong associ-
ation between MPH use and cancer risk in children, although their
ability to examine dose and duration of use or risk at specific sites
was limited by small numbers of participants. Furthermore, the
authors called for further study of MPH use and lymphocytic leu-
kemia risk to determine whether their results are caused by chance
alone. The limitations of the study include no clear description of
the representativeness of the sample, no base rate calculation of
cancer in the ADHD sample, and a shortage of detailed treatment
To our best knowledge, so far, the issue of an increased risk for
cancer under medications other than stimulants for ADHD has not
been raised in the literature. However, antidepressants are fre-
quently used as a second line drug for ADHD, and antipsychotics
are also frequently used by clinicians for this condition. For the
latter, it is known that, for example, taking clozapine entails an
eight times increased risk for acute myeloid leukemia (Nielsen and
Given the current limited epidemiological evidence on whether
or notthere is a significant association between MPH treatment and
cancer, the objective of the present study was to take a new look at
the issue based on large and representative data from nationwide
registers in Denmark. In this approach, data on treatment for
ADHD with psychoactive drugs other than MPH, including anti-
depressants and antipsychotics, and both dose and duration of
The study was approved by the Danish Data Protection Agency,
National Board of Health and Statistics Denmark. In Denmark, at
birth each individual is assigned a unique identification number
(CPR) which makes it possible to identify the individual across
various registers. The patients included in this study had to be
included in the Integrated Database for Labour Market Research
(IDA database) in order to extract birth year and gender informa-
tion. For the present study, we used data from four additional
Danish Psychiatric Central Register (DPCR).
base was used for the identification of all patients diagnosed with
ADHD in the period between 1994 and 2010. The time period was
disorders in the current International Statistical Classification of
Diseases and Related Health Problems, 10th Revision (ICD-10),
since1994. The definition of ADHD according to the Diagnostic and
Statistical Manual of Mental Disorders (DSM) used in the United
States, and hyperkinetic disorder according to the ICD are not fully
identical, with the latter being less inclusive than the former. In
clinical practice, clinicians in Denmark are asked to diagnose ac-
cording to ICD-10 criteria. However, over the years, they have in-
clined also toward using the term ‘‘ADHD,’’ including its DSM
criteria and rating scales that have been derived from these criteria.
on all individually prescribed medications. ADHD drugs in this
register are defined according to the World Health Organization
(WHO) Anatomic Therapeutic Chemical classification system.
The followingdrugs (anatomical therapeutic chemical [ATC] code)
were considered: methylphenidate (NA06BA04 and NA06BA11),
amphetamines (NA06BA01, NA06BA02, NA06BA03, N06BA06,
N06BA10 and N06BA12), other ADHD-specific drugs available
(pemoline NA06BA05, modafinil NA06BA07, fenozolone NA-
06BA08, and atomoxetine NA06BA09), antidepressants (NO6A),
and antipsychotics (NO5A).
Based on the entire observation period, the analyses considered
the following six subgroups: 1) MPH only treatment, 2) amphet-
amine only treatment, 3) other ADHD-specific drugs only treat-
ment, 4) antidepressants only treatment, 5) antipsychotics only
treatment, and 6) no medication. A large subgroup of patients re-
ceiving various combinations of drugs was not considered in the
analyses because of not allowing clear interpretations of drug ef-
fects. The prescription database does not contain data on drug
dosages but, rather, on the total number of sold defined daily
dosages (DDD) according to recommendations of the WHO Col-
laborating Centre of Drug Statistics Methodology (WHO Colla-
borating Centre for Drug Statistics Methodology 2011). The DDD
for MPH is 30mg. The DDD reflects the total exposure to drugs
over time without considering discontinuous periods of treatment.
Furthermore, we calculated the total time period for the various
drug treatments irrespective of doses and discontinuation of drug
This register contains information
The National Patient Registry (NPR).
are based on the ICD-10 classification codes. Both any cancer di-
classification were considered for analysis.
All cancer diagnoses
The Mortality Register.
causes of death of participants and was only used in the analysis
using survival methods. We followed the patients from either onset
of medication or birth (in the no medication subgroup) until either
cancer or death occurred. If none of these events occurred, the data
for these individuals were censored on December 31, 2010.
This register includes dates and
MEDICATION FOR ADHD AND CANCER209
deviations were computed for each intervention group. The times
before and after onset of drug treatment were compared by use of z-
statistics (Wilcoxon signed rank test). Frequencies of any cancer
diagnosis were calculated for the time before or after onset of drug
treatment for ADHD. Differences between frequencies of cancer in
these two groups and between frequencies of cancer in the after
compared using Fisher’s exact test. The impact of the various drug
interventions on any cancer was studied by using survival analysis
for time varying covariates, Kaplan–Meier statistics, and the log
rank test. Statistical analyses were performed by use of the Stata
statistical software version 11.
A total of n=21,186 patients with an ADHD diagnosis in the
years between 1994 and2010 coveredby the NPRwere suitablefor
the main analysis. We identified a total of n=15,319 patients who
received at least one prescription for MPH, amphetamines, other
ADHD drugs, antidepressants, or antipsychotics, or a combination
of drugs. In this sample, n=7418 (50.95%) received MPH only, 28
(0.19%) amphetamines only, 115 (0.79%) other ADHD-specific
drugs only, 704 (4.84%) antidepressants only, and 426 (2.93%)
antipsychotics only. A subgroup of n=6628 (43.27%) patients re-
ceived various combinations of these drugs over time and were
called the ‘‘mixed drugs subgroup.’’ and a total of n=5867 patients
with ADHD who had never received any medication served as a
control group. The final sample size used for the present analyses
was n=21,186 including 15,940 (75.24%) males and 5246
The duration of recommended drug doses in terms of DDD and
the total time period before and on treatment are shown in Table 1,
indicating that MPH was the major drug of choice in the observa-
commercial products available, and pharmacy prescription had to
be used. In a sizeable proportion of the sample, antidepressants and
antipsychotics also had been prescribed.
The various drug subgroups differed significantly in age at onset
of medication, with MPH and amphetamines prescribed in child-
hood, other ADHD-specific drugs in adolescence, and antidepres-
deviations of the years on DDDs were lower than the period after
onset of drug treatment. The mean period on drugs varied between
1.29 and 10.78 years for the five drug subgroups. With the excep-
tion of a reverse situation with the amphetamines, all observation
periods before onset of drug treatment were significantly longer
than the periods on the various drugs.
In the total sample, 68/21,186 patients (0.32%, or 1.80/10,000
patient-years) with a diagnosis of cancer were identified. There
were 29/21,186 patients (0.14%, or 1.27/10,000 patient-years) who
had received their diagnosis before any drug treatment. An addi-
tional 30/21,186 patients (0.14%, or 4.33/10,000 patient-years)
patients (0.04%, or 1.18/10,000 patient years) developed cancer
without ever having received any medication. The numbers and
rates of any cancer diagnoses in relation to onset of drug treatment
in the six subgroups with various drug interventions and in the
group without any medication are shown in the upper part of
Table 2. Only in the antipsychotics subgroup was the frequency of
any cancer significantly higher after than before the onset of drug
treatment. Furthermore, only the mixed subgroup had a signifi-
cantly higher cancer rate than the group without any medication.
However, the rates of cancer before and after onset of medication
did not differ significantly in the MPH subgroup.
The frequencies and percentages of the various subtypes of
cancer in relation to onset of drug treatment are shown in the lower
part of Table 2. Under all four conditions, malignant neoplasms of
lymphoid, hematopoietic, and related tissues were most frequent,
followed by neoplasms ofthe male orfemale genital organsand the
skin. There was no specific pattern of cancer diagnoses in the no
medication subgroup and there were no significant differences
when comparing the specific cancer frequencies before and after
onset of drug treatment, on the one hand, and after onset of drug
treatment and under no medication, on the other hand.
After removing the 28 and 115 patients who had only received
amphetamine or other ADHD- specific drugs from the cohort be-
cause of no observed cancer cases in these treatment groups, a
follow-up study was performed with the remaining 21,043 patients
by using Cox regression analysis adjusted for time varying cov-
ariates (i.e., treatment group, age at onset of medication, and du-
ration of treatment, i.e., years on DDD) and year of birth, and
stratified on sex. To study the particular drug risk on cancer, it was
defined that patients became at risk of cancer at birth in order to
have the same follow-up period for all participants. Only the MPH,
Table 1. Duration of Treatment with Defined Daily Doses and Observation
Period of Treatment in Various Drug Subgroups
Years on DDD Period before drug onset (years)aPeriod after drug onset (years)
Other ADHD-specific drugs
aKruskal–Wallis test: v2=3487.79, df=5, p= <0.05.
bComparisons of the periods before drugs and on drugs.
ADHD, attention-deficit/hyperactivity disorder; DDD, defined daily doses.
210STEINHAUSEN AND HELENIUS
no medication subgroups were included because the two other
subgroups (amphetamines or other ADHD-specific drugs) did not
develop any cancer after onset of drug treatment. This procedure
resulted in a sample of 21,043 patients. The treatment group of
reference was the no medication subgroup. The use of treatment
group as a time varying covariate ensured that each patient was in
the no medication subgroup until the date of onset of medication,
when he or she switched to MPH, antidepressant, antipsychotic, or
mixed treatment. In the same way, age at onset of medication and
duration of medication was set to 0 whenever an individual was not
receiving treatment. The findings show that none of the various
drug conditions represented a risk factor, whereas dose (re-
presented by years on DDD) and earlier birth (older age), were risk
factors for developing cancer (Table 3).
Figure 1 provides information coming from survival analysis
depicting the time from onset of drug treatment until cancer was
Table 2. Number (Percentages) of Patients with any Cancer in Various Drug Groups and Specification
of Cancer Diagnoses Before and After Medication and with no Medication
(A) Cancer before onset
of drug treatment (n=29)
(B) Cancer after onset
of drug treatment (n=30)
(C) No medication
Other ADHD-specific drugs
Lip, oral cavity, and pharynx
Respiratory and intrathoracic organs
Bone and articular cartilage
Mesothelial and soft tissue
Female genital organs
Male genital organs
Eye, brain, and other parts of central
Thyroid and other endocrine glands
Malignant neoplasm of ill-defined,
secondary, and unspecified sites
Malignant neoplasm, stated or presumed
to be primary, of lymphoid,
hematopoietic and related tissue
- 1 (3.33)
12 (41.38)8 (26.67) 1 (11.11)0.280.65
aFishers exact test comparing A and B.
bFishers exact test comparing B and C.
ADHD, attention-deficit/hyperactivity disorder.
Table 3. Hazard Ratios Associated with Various Drugs
in Relation to any Cancer Adjusted for Time
Varying Covariates and Stratified on Sex
Years on DDD
Year of birth
Age at medication
DDD, defined daily doses.
various drug subgroups.
Survival analysis findings (Kaplan–Meier curves) for
MEDICATION FOR ADHD AND CANCER211
subgroup with no medication. The log rank test was applied to test
the equality of the survivor functions across the subgroups, and the
finding indicated significant differences in the survival curves of
the various drug subgroups (v2=60.23, df=4, p<0.001).
treatment of ADHD with MPH or other drugs and cancer, based on
a large and representative nationwide dataset. A unique possibility
to link various register-based datasets in Denmark allowed specific
tests of the claim based on a small study with children that the
exposure to MPH might be associated with the development of
cancer (El-Zein et al. 2005). Data on various other drugs used for
the treatment of ADHD and from an unmedicated group of patients
with ADHD were analyzed for comparison. The mean DDD ob-
servation time of MPH exposure was 2.7 years. For antidepressant
or antipsychotic drug exposure, the mean DDD observation period
was shorter, whereas it was longer for the subgroup receiving
mixed drugs. Because ADHD is diagnosed predominantly in
childhood, most of the observation time took place during ado-
lescence and young adulthood.
The prevalence rates for any cancer and particularly under MPH
treatment were rather low. The original speculation for any po-
tentially causal association of MPH treatment and cancer because
of an increased rate of chromosomal abnormalities (El-Zein et al.
2005) did not receive support from the present findings for various
reasons. First, with an equal frequency of any cancer before and
after onsetofMPH treatment, therewasno indicationthatcancerin
ADHD patients might be the result of MPH treatment. However, it
has to be noted that the small number of cancer patients after onset
of MPH treatment was significantly higher than under the control
condition of ADHD without any medication ever.
Second, the direct comparisons of the various specific cancer rates
for patients before or after onset of drug treatment did also show
similar rates. All associations were chance findings. The rates were
highest but not significantly increased in neoplasms of the lymphoid,
hematopoietic, and related tissue. The single other epidemiological
study addressing the present research question found lymphocytic
leukemia to be most frequent in MPH-treated patients (Oestreicher
et al. 2007). However, this observation was not sufficiently con-
trolled, and may have been a chance finding, as the authors ac-
knowledge. In fact, the present findings do not point to a significant
association, because the rates for these specific cancers did not differ
significantlyamong thosediagnosed with cancerbeforeorafteronset
The finding of five patients who developed cancer after onset of
medication with antipsychotics may match the results of the register
study by Nielsenand Boysen (2010),indicating that clozapine hasan
increased risk for acute myeloid leukemia.
The risk model based on HR ratios with the no medication
subgroup as reference clearly showed that none of the drugs, in-
Therefore, there is another strong argument against any cancero-
genic effects of MPH in particular. It became quite evident that
were relevant. These findings were also controlled for effects
caused by varying time periods in drug exposure, duration of
treatment, and age at onset of treatment. Therefore, the two general
factors ofdoseandage,rather thananyspecificdrug,contributedto
the risk of developing cancer.
The dose effect could imply a generic effect of drug treatment,
or simply a greater power effect in this analysis, than for any one
drug. An effect caused by multiple drug use would be less likely,
because the mixed drug effects in the analysis were not associated
link between severity of ADHD and cancer because the dataset
was based on categorical ADHD diagnoses only and did not
contain a severity indicator. However, we controlled in our ana-
lyses for use of antipsychotics, excluding that this potential
marker of illness severity had a relevant effect on the significant
dose–response relationship. In addition, one needs to bear in mind
that this was only a post-hoc analysis, and that multiple unmea-
sured variables are related to the use of higher ADHD medication.
Therefore, as we did not find an overall effect of ADHD medi-
cations, it could also well be that the dose relationship found in
our post-hoc analyses is an artefact. Nevertheless, future studies
assessing the association between ADHD medications and cancer
risk should pay attention to potential dose effects of ADHD
medications, and include measures of disease severity and other
variables that could influence both dosing choices of ADHD
medications and cancer risk. Finally, the age effect on cancer risk
simply implies the greater probability of getting cancer with in-
There are various limitations of the present study. First, the
analyses were not based on actual doses of medication, but rather
on DDD. Second, despite the much longer observation periods than
in all previous small-sized studies, the observation time period of
the present study amounted only to a 17 year period of data col-
lection. Observation periods before drug treatment were signifi-
cantly longer than periods on drug treatment. Among the 15,319
patients receiving drug treatment, a large proportion of ADHD
patients were diagnosed only after the year 2005, so that the time to
developing cancer was rather short. However, one may argue that
cancer latencies may take much more time, as much as up to 20
years, so that more long-term epidemiological follow-up studies
may be warranted.
The reported findings indicate that neither MPH nor any other
drug commonly used in ADHD treatment has any specific and
significant association with cancer within the observed time frame.
However, longer observation periods of medicated patients with
ADHD with a specific focus on higher doses are clearly warranted.
This study found no significant association of medications used
for ADHD and cancer. Given the large numbers of individuals
diagnosed with ADHD and preferentially treated with medication,
this conclusion may reduce concerns about drug treatment among
the affected patients, their relatives, and the public at large.
H.-C. Steinhausen has worked as an advisor and speaker for the
following pharmaceutical companies: Janssen-Cilag, Eli Lilly,
Novartis, Medice, Shire, and UCB. In the past, he has received
unrestricted grants for postgraduate training courses or conferences
by Janssen-Cilag, Eli Lilly, Novartis, Medice, and Swedish Orphan
International. Within the last 4 years, he has not received any fi-
nancial support for drug studies. Dorte Helenius has no conflicts of
financial interest to disclose.
212STEINHAUSEN AND HELENIUS