Database processing for identification of
concomitant drug frequencies in a
forensic material positive for
antidepressant drugs
Niklas Björn, 2014-06-11
LiU−IMH−EX−14/01−SE
Master Thesis conducted at the department of Medical and Health Sciences, division
of Drug Research, Clinical Pharmacology
Supervisor: Margareta Reis, Department of Medical and Health Science, University
of Linköping
Examinator: Henrik Gréen, Department of Medical and Health Science, University of
Linköping
Department of Medical and Health Sciences
Linköping University
Avdelning, institution
Division, Department
Division of Drug Research, Clinical Pharmacology Department of Medical and Health Science Linköping University
URL för elektronisk version
ISBN
ISRN:
LiU−IMH−EX−14/01−SE
_________________________________________________________________
Serietitel och serienummer ISSN
Title of series, numbering ______________________________
Språk Language Svenska/Swedish Engelska/English ________________ Rapporttyp Report category Licentiatavhandling Examensarbete C-uppsats D-uppsats Övrig rapport _____________ Titel Title
Database processing for identification of concomitant drug frequencies in a forensic material
positive for antidepressant drugs
Författare
Author
Niklas Björn
Nyckelord
Keywords
TCA, TeCA, SSRI, SNRI, MAOI, antidepressant drug, drug interaction, intoxication
Sammanfattning
Abstract
This article presents a study conducted on data containing drug concentrations. The data was obtained
from femoral venous blood samples collected at medico legal autopsies in Sweden. Cases positive for
antidepressant drugs were scrutinized and divided in to two groups for 15 antidepressant drugs: B-cases,
where the cause of death was intoxication with more than one drug detected in the blood sample. C-cases,
where the cause of death was NOT intoxication and at least one drug (the antidepressant) was detected in
the blood sample. This data was then processed to find frequencies of concomitant drugs taken together
with the antidepressant drugs. Frequencies of the most common concomitant drugs were then compared
between B-cases and C-cases for each antidepressant drug. This revealed that the drugs
dextropropoxyphene, ethanol, codeine, flunitrazepam, paracetamol, propiomazine and alimemazine were
signifcantly more common as concomitant drugs in B-cases (intoxications) than in C-cases
(non-intoxications). With regards to unknown interactions the most interesting combinations were:
Propiomazine with mirtazapine, venlafaxine, citalopram or fluoxetine; Paracetamol with paroxetine;
Flunitrazepam with mirtazapine, venlafaxine or citalopram; Codeine with mirtazapine or sertraline. These
combinations should be further investigated
.Datum
Date
Table of Contents
1. Abstract ... 1
2. List of abbreviations ... 1
3. Introduction ... 2
3.1.
Background ... 2
3.1.1.
Assignment ... 2
3.1.2.
Data... 2
3.1.3.
Antidepressant drugs... 3
3.1.4.
Drug interactions ... 4
3.2.
Purpose ... 4
3.3.
Clinical relevance... 4
4. Process ... 4
5. Methods ... 5
5.1.
Extract data ... 6
5.2.
Parent drug and metabolite ... 7
5.3.
Frequencies ... 8
5.4.
Statistical analysis ... 9
5.4.1.
Contingency tables ... 9
5.4.2.
Fisher´s exact test ... 9
5.4.3.
Odds Ratio ... 10
5.4.4.
Relative Risk ... 10
5.4.5.
Correction for multiple comparisons ... 11
5.4.6.
Comparisons of concentrations ... 11
5.5.
Known drug interactions... 12
6. Result ... 12
6.1.
All cases ... 12
6.2.
Index-antidepressants and concomitant drugs ... 12
6.2.1.
Amitriptyline ... 13
6.2.2.
Citalopram ... 14
6.2.3.
Clomipramine ... 14
6.2.4.
Fluoxetine ... 15
6.2.6.
Mirtazapine ... 16
6.2.7.
Paroxetine ... 16
6.2.8.
Sertraline ... 17
6.2.9.
Venlafaxine ... 17
7. Discussion ... 18
8. Conclusions ... 20
9. Acknowledgments... 20
10. References ... 21
Appendix I – Known drug interactions ... 23
Appendix II – Relative risk ... 27
Appendix III – Frequency tables ... 44
1
1. Abstract
This article presents a study conducted on data containing drug concentrations. The data
was obtained from femoral venous blood samples collected at medico legal autopsies in
Sweden. Cases positive for antidepressant drugs were scrutinized and divided in to two
groups for 15 antidepressant drugs: B-cases, where the cause of death was intoxication with
more than one drug detected in the blood sample. C-cases, where the cause of death was
NOT intoxication and at least one drug (the antidepressant) was detected in the blood
sample. This data was then processed to find frequencies of concomitant drugs taken
together with the antidepressant drugs. Frequencies of the most common concomitant drugs
were then compared between B-cases and C-cases for each antidepressant drug. This
revealed that the drugs dextropropoxyphene, ethanol, codeine, flunitrazepam, paracetamol,
propiomazine and alimemazine were signifcantly more common as concomitant drugs in
B-cases (intoxications) than in C-B-cases (non-intoxications). With regards to unknown
interactions the most interesting combinations were: Propiomazine with mirtazapine,
venlafaxine, citalopram or fluoxetine; Paracetamol with paroxetine; Flunitrazepam with
mirtazapine, venlafaxine or citalopram; Codeine with mirtazapine or sertraline. These
combinations should be further investigated.
Keywords: TCA, TeCA, SSRI, SNRI, MAOI, antidepressant drug, drug interaction,
intoxication
2. List of abbreviations
TCA : Tricyclic antidepressant
TeCA : Tetracyclic antidepressants
SSRI : Selective-serotonin reuptake inhibitors
SNRI : Serotonin/noradrenaline reuptake inhibitors
MAOI : Monoamine oxidase inhibitors
ICD-9 : International Classification of Diseases, Ninth Revision
OR : Odds ratio
RR : Relative risk
CI : Confidence interval
2
3. Introduction
3.1. Background
Access to parts of a nationwide database for forensic pathology and forensic toxicology was
available at the department of clinical pharmacology, Linköping University, Linköping. This
database contained drug concentration data obtained from femoral venous blood samples
collected at medico legal autopsies in Sweden. The concentration data is accompanied with,
among other information, the cause of death and the manner of death [1]. All handling and
processing of the blood samples of medico legal autopsies are strictly standardized at all
forensic pathology units according to guidelines set by the Swedish National Board of
Forensic Medicine. During the period January 1992 to June 2005, 55 042 samples were
screened for substances, including antidepressant drugs. The data from the samples positive
for antidepressant drugs formed the basis for the article Reis et al. [2], and constitutes the
base for the present project.
3.1.1. Assignment
The aim was to process the antidepressant concentration data for further findings. More
specifically, to investigate frequencies of concomitant drugs and combination of concomitant
drugs used together with different antidepressant drugs. Subsequently, concentrations of
antidepressant drugs were scrutinized for concentration deviations that could be linked to
concomitant drugs detected in the samples. This might lead to proposals of unknown drug
interactions that should be further studied or strengthen the evidence for known drug
interactions.
3.1.2. Data
The data used in this project included all cases with a positive detection of one of 15 different
antidepressant drugs. These cases were divided into 1 of 3 distinct groups for each
antidepressant drug detected in the blood sample. This division was made in the former
study [2] were every single case was systematically investigated and divided in to these
groups. All in accordance with the Swedish version of International Classification of
Diseases, Ninth Revision (ICD-9)[3], together with the death diagnosis made by the
pathologist. Further, cases that gave rise to confusion or uncertainties as well as
unreasonably high or low concentrations of the antidepressant were excluded. The three
groups that were formed as in the previous study were denominated A-cases, B-cases, and
C-cases:
A-cases: cases dead by intoxication, with only one single drug detected in the blood sample.
Influence by alcohol was allowed up to 0.1%.
B-cases: cases dead by intoxication, with more than one drug detected in the blood sample.
Influence by alcohol was tolerated at all concentrations.
C-cases: cases where the cause of death was NOT intoxication and at least one drug was
3
This means that a case can only be present in one A-case group. However, a case with more
than one antidepressant drug present could be represented in B-case groups for different
antidepressant drugs or in C-case groups for different antidepressant drugs.
3.1.3. Antidepressant drugs
The most common theory of depressions is the monoamine theory, proposed by Schildkraut
in 1965. The theory states that a depression is caused by a deficit of the monoamine
neurotransmitters, serotonin and noradrenaline, at specific sites in the brain. The base for
this theory is derived from the clinical effect of the drugs used to treat depressions.
Antidepressant drugs induce in various different ways an increase of the amount of
neurotransmitters in the synaptic cleft between two neurons. This is supposed to improve
the mood of the patient. [4]
There are a many types of antidepressant drugs, and below comes a short introduction of the
types of antidepressant drugs relevant to this project and brief theories behind the drug
actions.
Tricyclic antidepressants (TCAs) increase the amount of neurotransmitters in the synaptic
cleft by inhibiting the reuptake of both serotonin and noradrenaline from the synaptic cleft.
These were the first generation of antidepressant drugs [4]. The antidepressant effect is good
but side effects are common and sometimes serious/life-threatening. Even today, the more
severe the depression is, the more likely the patient is to be treated with TCAs. Seizures and
ventricular arrhythmia account for the mortality associated with TCA overdoses [5].
Selective-serotonin reuptake inhibitors (SSRIs) are antidepressants that are more prone to
inhibit serotonin reuptake than noradrenaline reuptake from the synaptic cleft. These
antidepressants have fewer side effects than TCAs and are harder to overdose. [4, 6]
Serotonin/noradrenaline reuptake inhibitors (SNRIs) are more selective for inhibition of
noradrenaline reuptake. SNRIs are considered safer in overdose than TCAs [4].
Monoamine oxidase inhibitors (MAOIs) inhibit the enzyme monoamine oxidase which
regulates serotonin and noradrenaline concentrations. MAOIs are used less than other
antidepressants due to their adverse effects and interactions [4, 7]. MAOIs should not be
used in combination with other serotonin enhancing drugs, e.g. other antidepressant drugs,
because of the increased risk for serotonin syndrome [8]. Serotonin syndrome is a potentially
life-threatening syndrome that can occur by therapeutic use of serotonergic drugs alone or
by drug interaction between two serotonergic drugs [9].
Tetracyclic antidepressants (TeCAs) are a group of antidepressants named after their
chemical structure in the same manner as TCAs. Mianserin and Mirtazapine act by blocking
adrenergic and serotonergic receptors which leads to an enhanced release of the
neurotransmitters serotonin and noradrenaline. Maprotiline is another TeCA it works as a
noradrenalin reuptake inhibitor. [4]
4
Antidepressants can be used as treatments for other conditions than depressions for example
anxiety, neuropathic pain, anorexia, bulimia and social phobia [4, 10]. The different
antidepressant drugs display a wide variety of side effects and the risk for life threatening
effects differ [4, 8].
3.1.4. Drug interactions
Drugs in general can interact with other drugs and the more drugs a patient has taken the
higher is the risk for unwanted side effects due to interactions. There are two general types
of drug interactions [4]:
Pharmacodynamic interaction: the effect of a drug is changed without a change in
concentration of that drug.
Pharmacokinetic interaction: the concentration of a drug is changed.
Drug interactions can lead to intoxications and even to death. In this study there was
information about the drugs that were taken prior to death, evidenced by post mortal blood
concentrations. This can be used to propose combinations of drugs that might have
underlying drug interactions.
3.2. Purpose
The main purpose of this project was to compare previously defined B-cases and C-cases to
find frequencies of concomitant drugs and combinations of concomitant drugs that could be
linked to intoxications with antidepressant drugs.
3.3. Clinical relevance
Depression is a growing health burden estimated to be one of the leading causes of mental
and health disability by 2030 [11]. Information about drugs used together with
antidepressant drugs is highly relevant for the safety of drug users, and for the subscribing
doctors. This study gives insight into poly-drug use and the prevalence of different drugs
used in combination with antidepressant drugs. The dangers associated with poly-drug use
is hard to interpret and common in intoxications [12]. Also the data in this project give
information from doses higher than therapeutic levels [2]. High toxic doses are not studied in
human trials. Drugs considered safe in combination might in high doses show different
tendencies. With all this in mind it is easy to understand why extended knowledge about
antidepressant drugs and their drug interactions is essential to future healthcare.
4. Process
The project started with the development of a structured timetable, see figure 1. This
timetable was followed without any major changes.
5
Figure 1: Timetable for the project.
The first part of the project was to get familiar with the data. This was followed by
anonymization of the data so that work could be conducted without the possibility of leaking
classified information. After this all groups where extracted, divided and stored. The number
of cases in each group were compared with the number of cases in the previous article [2].
Some small differences were found and those were linked to human errors made in the
former study. This was thoroughly discussed with the supervisor.
Concomitant drug frequencies and concomitant drug combination frequencies were
calculated. With this in hand statistical analyzes could be conducted. Lots of focus was put
on making sure the numbers were correct and how to do the statistical analyzes. A couple of
times tasks performed earlier had to be remade because of initial programming errors. All in
all the project went fine and always forward. Regular meetings were held with the
supervisor for asking questions and to check in with direction to continue the project.
The halftime meeting was held week 13, to follow up what had been done and how to direct
future work. No major changes were made and the project could continue as planned. In the
end of the project time was only spent on the report. However, sometimes earlier steps in the
timetable had to remade to obtain more or different information.
5. Methods
MATLAB
®a high-level language and interactive environment for numerical computation,
visualization, and programming was used for this project. Functions and scripts were
specifically designed for all different parts handling the data, both for extraction/handling of
data and for calculations conducted on the extracted data.
6
5.1. Extract data
Per definition the A-cases have no other substances present and are therefore not interesting
when researching drug combinations and drug interactions. Hence, only B-cases and C-cases
were extracted for the 15 different antidepressant drugs; i.e. 15 groups of B-cases and 15
groups of C-cases. One B-case group and one C-case group will have the same
antidepressant drug in common, this antidepressant drug will be referred to as the
index-antidepressant. The total number of cases in each group and the 15 antidepressant
drugs are found in Table 1. The information of concomitant drugs detected in each case was
also extracted. Classification codes from ICD-9 [3] for the cause of death and the manner of
death were used to exclude unreliable cases, for codes that lead to exclusion of C-cases see
Table 2.
Table 1: Shows the different antidepressant drugs, type of antidepressant, number of B-cases, number of C-cases, total number of cases for each antidepressant, total number, number of unique B-cases and number of unique C-cases.
7
Table 2: Classification codes for the manner of death and the cause of death that lead to exclusion of C-cases, where death by incapacitation and intoxication could not be ruled out. Table by courtesy of Reis et al. [2].
The index-antidepressants were also combined into two total groups; all B-cases and all
C-cases. This serves a purpose by giving a more general view over concomitant drug use in
cases positive for antidepressant drugs. Cases that were present in more than one of the
index-antidepressant groups that were combined were only used once in the new group. For
these two groups the concentration of the index-antidepressants is not interesting, only the
presence of them is relevant.
5.2. Parent drug and metabolite
Drugs are metabolized to different extent. In this study parent drugs or main metabolites
were viewed as an indication of the parent drug in question. The indication of a drug was
named by the parent drug, but includes the following three possible findings in the blood
sample: parent drug, metabolite or parent drug and metabolite. Parent drugs and their
respective main metabolites are listed in Table 3. The antidepressant amitriptyline has an
active metabolite nortriptyline which is sold as an antidepressant drug by itself. Therefore,
these two were not seen as the same drug.
8
Table 3: Shows index-antidepressants, the most common concomitant drugs and their respective main metabolites.
For the index-antidepressants the parent drug had to be present in the blood sample for a
case to be included in that index-antidepressants group. Concentrations were not used for
the concomitant drugs, they were only seen as present or not present. This enabled both
inclusion of cases were a concomitant drug was indicated but the concentration was not
measurable, and the possibility of combining parent drugs and metabolites.
5.3. Frequencies
MATLAB
®scripts were made to go through the data case by case for each
index-antidepressants B-case and C-case group. While doing this all detected substances
used by the cases in the group were noted and the frequencies of them were calculated.
Simultaneously, different double and triple combinations of concomitant substances were
detected and the frequencies of these combinations were calculated. The result from this is
shown in frequency tables, where the number of cases in the group, the total number of
different substances in the group and the number of cases with 1, 2 or 3 substances in
addition to the index-antidepressant (or in addition to all index-antidepressants in the two
combined groups). This gives an idea of how common it is to mix drugs. Median
9
median concentration was also calculated for the cases with at least one concomitant
substance and for the cases without any concomitant substances. These two groups were
compared to see if there was any significant difference between the concentrations in the two
groups, see 5.4.6. Comparisons of concentrations.
When analyzing the frequencies of concomitant drugs the 15 most common substances were
considered. When combinations of concomitant drugs used together with the
index-antidepressant were analyzed the 20 most common combinations were considered.
The combinations are either consisting of 2 or 3 drugs in addition to the
index-antidepressant. Combinations of more than three drugs were possible but very rare,
therefore these were not analyzed.
5.4. Statistical analysis
Differences with a p < 0.05 were α = 0.05 was considered significant.
5.4.1. Contingency tables
Statistical analyses were made to find concomitant drugs that were statistically more
common in B-cases than in C-cases or statistically more common in C-cases than in B-cases.
Contingency tables were formed for all concomitant drugs that were present in at least 10%
of the B-cases or C-cases. Figure 2 shows both a common contingency table and an example
of how they were formed in this project. Contingency tables are not displayed but the
numbers for a, b, c and d are shown were needed.
Figure 2: Contingency tables, on the left a common example and to the right an example of a contingency table in this project.
a: the number of B-cases where the concomitant drug was present. b: the number of C-cases where the concomitant drug was present. c: the number of B-cases where the concomitant drug was NOT present. d: the number of C-cases where the concomitant the drug was NOT present.
5.4.2. Fisher´s exact test
Fisher’s exact test was conducted on all contingency tables for investigating if there were
differences between the variables in the rows and columns of the contingency tables. Fisher’s
exact test is suitable when small numbers could be present in any of the rows or columns.
Other methods like a Chi squared test would have to be corrected if numbers are small, this
is not the case with Fisher’s exact test. [13]
Corrections for multiple tests were performed with the Benjamini-Hochberg procedure, see
5.4.5. Correction for multiple comparisons.
10
5.4.3. Odds Ratio
This project is a case-control study and from the contingency tables odds ratios (OR) can be
calculated to give information about if the odds of an outcome is higher or lower if an
exposure is present [14]. The formula for calculating the odds ratio and its confidence
interval is given below [13].
The odds ratio is given by:
The standard error (SE) of the log odds ratio is given by:
√
The confidence interval (CI) is given by:
( )
note that N(x) is a value taken from the normal distribution
An OR = 1 means that the odds of the outcome (intoxication) is the same for both the exposed
(concomitant drug present) and the non-exposed (concomitant drug NOT present). For the
OR to be statistically significant the CI should not contain 1. If the OR > 1, then the odds for
the outcome is higher for the exposed group. If the OR < 1, then the odds for the outcome is
lower for the exposed group, a somewhat protective effect is seen. [14]
5.4.4. Relative Risk
In a case-control study should the relative risk (RR) not be directly calculated and often it is
completely wrong to calculate it [15]. However there are methods for deriving the RR from
the OR (and the corresponding CI). This can be done if control groups were properly derived
and by using some auxiliary information. The method used for this can be seen below [16].
Probability of diseased among the exposed:
Conversion of odds ratio to relative risk:
This conversion has flaws and some argue that it is not proper to do [17]. This conversion
might give values that still not should be viewed as risks. I argue that it still is appropriate to
do because a RR is much easier to understand compared to an OR [15]. The interesting is if
the RR and the associated CI is separated from 1 not how high or low the actual RR values
are.
An RR = 1 means that the risk of the outcome (intoxication) is the same for both the exposed
(concomitant drug present) and the non-exposed (concomitant drug NOT present). For the
RR to be statistically significant the CI should not contain 1. If the RR > 1, the risk for the
outcome is higher for the exposed group. If the RR < 1, then the risk for the outcome is lower
for the exposed group, a somewhat protective effect is seen. [14]
Specific explanation for the present project: RR > 1 means that death by intoxication is more
likely if the index-antidepressant is taken in combination with the concomitant drug in
question, compared to when the concomitant drug is not taken. An RR < 1 indicates the
11
opposite, that death by intoxication is less likely if the index-antidepressant is taken in
combination with the concomitant drug in question, compared to when the concomitant
drug is not taken.
5.4.5. Correction for multiple comparisons
Correction for multiple tests was done with the Benjamini-Hochberg procedure. The
procedure was conducted separately on all index-antidepressants or group of
index-antidepressants studied. The p-values from Fisher’s exact test on the most common
concomitant drugs were corrected. This controls the amount of type I errors (false
discoveries). [18]
Benjamini-Hochberg procedure, as explained in:
For n tests, rank corresponding p-values in ascending order , p
(1)< … < p
(n)
Let k be the largest i for which
is true
Tests corresponding to p
(1)… p
(k)can be considered significant at level α
Other tests possibly with p < α but with p > p
(k)are
not significant
In tables the p-value from Fisher´s exact test has to be lower than the value from
Benjamini-Hochberg procedure. This correction applies to the OR/RR results directly because
they were calculated from the same data. In RR plots there will be a black horizontal line, if a
confidence interval is beneath this line in the plot it is significant, otherwise it is not
significant even if the confidence interval does not include 1. So for a drug to be significantly
more common in one group both the p-value from Fisher´s exact test has to be lower than the
value from the Benjamini-Hochberg procedure and the OR/RR confidence interval has to be
separated from 1.
5.4.6. Comparisons of concentrations
Comparisons of concentrations between two groups were conducted with Mann-Whitney
U-test, a non-parametric test for unevenly distributed data.
Concentration comparisons were only made on the index-antidepressants.
First comparisons of C-cases with no concomitant drugs were compared with C-cases that
have at least one concomitant drug.
Secondly comparisons were made within B-cases and C–cases for drugs that were
significantly more common (Fisher´s exact test was significant) in B-cases or C-cases. For
finding differences in the concentration between cases with a certain concomitant drug
compared to cases without that concomitant drug. For example; for index-antidepressant X,
the concomitant drug Y was significantly more common in B-cases than C-cases. Then a
comparison of index-antidepressant X ´s concentration between B-cases with drug Y and
B-cases without drug Y was conducted. Also a comparison of index-antidepressant X ´s
concentration between C-cases with drug Y and C-cases without drug Y was conducted.
12
No corrections for multiple comparisons are made for the statistical tests on the
concentration differences.
5.5. Known drug interactions
The number of known drug-drug interactions is extensive and growing. In Appendix I
recognized interactions with index-antidepressants are listed. This was done manually by
searching SFINX (Swedish, Finnish, Interaction X-referencing) [19], a drug-drug interaction
database. This database is under constant development and these interactions could be
found April 4, 2014.
6. Result
All relative risk plots are displayed in appendix II. The most common combinations of
concomitant drugs are not shown in the results, however they can be found along with all
information from the frequency tables in appendix III. Further, results from the statistical
analysis including number of cases in each group (a, b, c, d), OR, RR, CI, p-values,
Benjamini-Hochberg correction and consentration comparisons are all listed in appendix IV.
6.1. All cases
All B-cases consisted of 749 cases and a total of 111 different drugs in addition to the
index-antidepressants were found. 736 cases had positive detections of drugs in addition to
the index-antidepressants. The 5 most common concomitant drugs were ethanol (found in
47% of the cases), paracetamol (32%), dextropropoxyphene (30%), propiomazine (24%) and
diazepam (16%).
All C-cases consisted of 1654 cases and a total of 89 different drugs in addition to the
index-antidepressants were found. 922 cases had positive detections of drugs in addition to
the index-antidepressants. The 5 most common concomitant drugs were ethanol (27%),
diazepam (22%), paracetamol (18%), zopiclone (15%) and propiomazine (13%).
Comparisons of frequencies of concomitant drugs between all B-cases and all C-cases is
displayed a relative risk plot (figure A, appendix II). This plot, p-values and confidence
intervals reveals that the drugs dextropropoxyphene, ethanol, codeine, flunitrazepam,
paracetamol, propiomazine, alimemazine were signifcantly more common in B-cases
(intoxications) than in C-cases (non-intoxications). Dextropropoxyphene was more common
than ethanol, paracetamol, propiomazine and alimemazine. Codein was more common than
than paracetamol, propiomazine and alimemazine. Furhtermore, flunitrazepam was more
common than alimemazine.
6.2. Index-antidepressants and concomitant drugs
13
Table 4: Lists the 5 most common concomitant drugs in ascending order for each index-antidepressant. For the combined group of all B-cases the 10 most common concomitant drugs are listed. The index-antidepressants are listed in no particular order.
Further results for the index-antidepressants fluvoxamine, imipramine, maprotiline,
moclobemide, nortriptyline and trimipramine will not be displayed, due to too few cases.
However, all the results for these index-antidepressants can be found in the appendices.
6.2.1. Amitriptyline
Amitriptyline B-cases consisted of 119 cases and a total of 71 different drugs in addition to
the index-antidepressant were found. The 5 most common concomitant drugs were ethanol
(56%), paracetamol (28%), dextropropoxyphene (25%), diazepam (23%) and flunitrazepam
(18%).
Amitriptyline C-cases consisted of 89 cases and a total of 39 different drugs in addition to the
index-antidepressant were found. 55 cases had positive detections of concomitant drugs. The
5 most common concomitant drugs were diazepam (33%), paracetamol (22%), ethanol (20%),
dextropropoxyphene (16%) and carbamazepine (13%). No concentration differences could be
seen (p = 0.36) between C-cases with concomitant drugs in addition to the
14
Comparisons of frequencies of concomitant drugs between amitriptyline B-cases and C-cases
is displayed in the relative risk plot (figure B, appendix II). Ethanol was significantly more
common in B-cases than in C-cases. No changes in the concentration of amitriptyline could
be linked to ethanol.
6.2.2. Citalopram
Citalopram B-cases consisted of 243 cases and a total of 85 different drugs in addition to the
index-antidepressant were found. The 5 most common concomitant drugs were ethanol
(48%), dextropropoxyphene (36%), paracetamol (31%), propiomazine (26%) and diazepam
(23%).
Citalopram C-cases consisted of 629 cases and a total of 78 different drugs in addition to the
index-antidepressant were found. 363 cases had positive detections of concomitant drugs.
The 5 most common concomitant drugs were ethanol (25%), paracetamol (19%), diazepam
(16%), propiomazine (14%) and zopiclone (13%). Concentration differences could be seen (p
= 0.03) between C-cases with concomitant drugs compared to C-cases without any
concomitant drugs. Median concentration of citalopram was significantly higher for C-cases
with concomitant drugs.
Comparisons of frequencies of concomitant drugs between citalopram B-cases and C-cases is
displayed in the relative risk plot (figure C, appendix II). Dextropropoxyphene, ethanol,
levomepromazine, propiomazine, flunitrazepam, paracetamol, diazepam and alimemazine
were all significantly more common in B-cases than in C-cases. Differences in concentration
of citalopram for cases with compared to without these drugs could be found for 3 of them:
Dextropropoxyphene; B-cases had no differences (p = 0.09), C-cases had higher citalopram
concentrations for cases with dextropropoxyphene (p = 0.01). Propiomazine; B-cases had no
differences (p = 0.16), C-cases had higher citalopram concentrations for cases with
propiomazine (p = 0.01). Flunitrazepam; B-cases had no differences (p = 0.35), C-cases had
higher citalopram concentrations for cases with flunitrazepam (p = 0.00). The plot also
reveles that dextropropoxyphene was more common than ethanol, propiomazine,
flunitrazepam, paracetamol, diazepam and alimemazine. Levomepromazine was more
common than paracetamol, diazepam and alimemazine.
6.2.3. Clomipramine
Clomipramine B-cases consisted of 107 cases and a total of 76 different drugs in addition to
the index-antidepressant were found. The 5 most common concomitant drugs were ethanol
(40%), paracetamol (31%), dextropropoxyphene (27%) diazepam (23%) and flunitrazepam
(14%).
Clomipramine C-cases consisted of 141 cases and a total of 51 different drugs in addition to
the index-antidepressant were found. 84 cases had positive detections of concomitant drugs.
The 5 most common concomitant drugs were diazepam (29%), ethanol (27%), alimemazine
(13%), paracetamol (12%) and dextropropoxyphene (7%). No concentration differences could
15
be seen (p = 0.37) between C-cases with concomitant drugs compared to C-cases without any
concomitant drugs.
Comparisons of frequencies of concomitant drugs between clomipramine B-cases and
C-cases is displayed in the relative risk plot (figure D, appendix II). Dextropropoxyphene
and paracetamol was significantly more common in B-cases than in C-cases. No changes in
the concentration of clomipramine could be linked to these drugs.
6.2.4. Fluoxetine
Fluoxetine B-cases consisted of 36 cases and a total of 50 different drugs in addition to the
index-antidepressant were found. The 5 most common concomitant drugs were diazepam
(44%), ethanol (42%), dextropropoxyphene (39%) propiomazine (31%) and paracetamol
(28%).
Fluoxetine C-cases consisted of 53 cases and a total of 37 different drugs in addition to the
index-antidepressant were found. 42 cases had positive detections of concomitant drugs. The
5 most common concomitant drugs were ethanol (45%), diazepam (31%), paracetamol (14%),
zopiclone (10%) and alimemazine (7%). No concentration differences could be seen (p = 0.81)
between C-cases with concomitant drugs compared to C-cases without any concomitant
drugs.
Comparisons of frequencies of concomitant drugs between fluoxetine B-cases and C-cases is
displayed in the relative risk plot (figure E, appendix II). Dextropropoxyphene and
propiomazine were significantly more common in B-cases than in C-cases. Differences in
concentration of fluoxetine for cases with compared to without these drugs could be found
for 1 of them: Dextropropoxyphene; B-cases had lower fluoxetine concentrations for cases
with dextropropoxyphene (p = 0.02), C-cases had no differences (p = 0.112).
6.2.5. Mianserin
Mianserin B-cases consisted of 28 cases and a total of 40 different drugs in addition to the
index-antidepressant were found. The 5 most common concomitant drugs were ethanol
(43%), paracetamol (36%), citalopram (25%) propiomazine (21%) and dextropropoxyphene
(18%).
Mianserin C-cases consisted of 156 cases and a total of 50 different drugs in addition to the
index-antidepressant were found. 108 cases had positive detections of concomitant drugs.
The 5 most common concomitant drugs were citalopram (21%), ethanol (16%), diazepam
(14%), sertraline (14%) and paracetamol (13%). No concentration differences could be seen (p
= 0.39) between C-cases with concomitant drugs compared to C-cases without any
concomitant drugs.
Comparisons of frequencies of concomitant drugs between mianserin B-cases and C-cases is
displayed in the relative risk plot (figure I, appendix II). Levomepromazine, ethanol,
16
common in B-cases than in C-cases. No changes in the concentration of mianserin could be
linked to these drugs.
6.2.6. Mirtazapine
Mirtazapine B-cases consisted of 67 cases and a total of 60 different drugs in addition to the
index-antidepressant were found. The 5 most common concomitant drugs were ethanol
(49%), propiomazine (34%), zopiclone (34%), dextropropoxyphene (33%) and diazepam
(30%).
Mirtazapine C-cases consisted of 237 cases and a total of 48 different drugs in addition to the
index-antidepressant were found. 162 cases had positive detections of concomitant drugs.
The 5 most common concomitant drugs were zopiclone (22%), diazepam (19%), citalopram
(14%), ethanol (14%) and paracetamol (14%). Concentration differences could be seen (p =
0.01) between C-cases with concomitant drugs compared to C-cases without any
concomitant drugs. Median concentration of mirtazapine was significantly higher for C-cases
with concomitant drugs.
Comparisons of frequencies of concomitant drugs between mirtazapine B-cases and C-cases
is displayed in the relative risk plot (figure J, appendix II). Dextropropoxyphene, ethanol,
codeine, morphine, propiomazine, flunitrazepam, paracetamol and alimemazine were
significantly more common in B-cases than in C-cases. Differences in concentration of
mirtazapine for cases with compared to without these concomitant drugs could be found for
2 of them: Dextropropoxyphene; B-cases had lower mirtazapine concentrations for cases
with dextropropoxyphene (p = 0.01), C-cases had no differences (p = 0.92). Propiomazine;
B-cases had no differences (p = 0.67), C-cases had higher mirtazapine concentrations for cases
with propiomazine (p = 0.00). The plot also show that dextropropoxyphene was more
common than alimemazine.
6.2.7. Paroxetine
Paroxetine B-cases consisted of 31 cases and a total of 41 different drugs in addition to the
index-antidepressant were found. The 5 most common concomitant drugs were paracetamol
(48%), dextropropoxyphene (36%), ethanol (32%), propiomazine (29%) and alprazolam
(16%).
Paroxetine C-cases consisted of 49 cases and a total of 24 different drugs in addition to the
index-antidepressant were found. 34 cases had positive detections of concomitant drugs. The
5 most common concomitant drugs were ethanol (29%), mianserin (21%), propiomazine
(18%), zopiclone (15%) and alprazolam (12%). No concentration differences could be seen (p
= 0.65) between C-cases with concomitant drugs compared to C-cases without any
concomitant drugs.
Comparisons of frequencies of concomitant drugs between paroxetine B-cases and C-cases is
displayed in the relative risk plot (figure M, appendix II). Paracetamol and
17
dextropropoxyphene were significantly more common in B-cases than in C-cases. No
changes in the concentration of paroxetine could be linked to these drugs.
6.2.8. Sertraline
Sertraline B-cases consisted of 46 cases and a total of 52 different drugs in addition to the
index-antidepressant were found. The 5 most common concomitant drugs were ethanol
(41%), dextropropoxyphene (26%), paracetamol (26%), propiomazine (22%) and citalopram
(17%).
Sertraline C-cases consisted of 196 cases and a total of 40 different drugs in addition to the
index-antidepressant were found. 124 cases had positive detections of concomitant drugs.
The 5 most common concomitant drugs were ethanol (27%), diazepam (19%), zopiclone
(18%), paracetamol (15%) and mirtazapine (13%). No concentration differences could be seen
(p = 0.277) between C-cases with concomitant drugs compared to C-cases without any
concomitant drugs.
Comparisons of frequencies of concomitant drugs between sertraline B-cases and C-cases is
displayed in the relative risk plot (figure N, appendix II). Codeine, dextropropoxyphene,
citalopram, caffeine and zolpidem were significantly more common in B-cases than in
C-cases. Differences in concentration of sertraline for cases with compared to without these
concomitant drugs could be found for 2 of them: Citalopram; B-cases had higher sertraline
concentrations for cases with citalopram (p = 0.03), C-cases had no differences (p = 0.42).
Caffeine; B-cases had lower sertraline concentrations for cases with caffeine (p = 0.05),
C-cases had no differences (p = 0.26).
6.2.9. Venlafaxine
Venlafaxine B-cases consisted of 54 cases and a total of 51 different drugs in addition to the
index-antidepressant were found. The 5 most common concomitant drugs were
propiomazine (41%), paracetamol (33%), ethanol (31%), zopiclone (31%) and alimemazine
(30%).
Venlafaxine C-cases consisted of 142 cases and a total of 37 different drugs in addition to the
index-antidepressant were found. 86 cases had positive detections of concomitant drugs. The
5 most common concomitant drugs were ethanol (29%), diazepam (21%), paracetamol (16%),
zopiclone (16%) and propiomazine (13%). No concentration differences could be seen (p =
0.38) between C-cases with concomitant drugs compared to C-cases without any
concomitant drugs.
Comparisons of frequencies of concomitant drugs between venlafaxine B-cases and C-cases
is displayed in the relative risk plot (figure P, appendix II). Dextropropoxyphene,
propiomazine, nortriptyline, codeine, alimemazine, amitriptyline, flunitrazepam, citalopram
and paracetamol were significantly more common in B-cases than in C-cases. Differences in
concentration of venlafaxine for cases with compared to without these concomitant drugs
18
could be found for 3 of them: Propiomazine; B-cases had no differences (p = 0.20), C-cases
had higher venlafaxine concentrations for cases with propiomazine (p = 0.01). Codeine;
B-cases had lower venlafaxine concentrations for cases with codeine (p = 0.05), C-cases had
no differences (p = 0.19). Alimemazine; B-cases had no differences (p = 0.33), C-cases had
higher venlafaxine concentrations for cases with alimemazine (p = 0.01).
7. Discussion
Table 5 presents a compilation of all concomitant drugs that were significantly more
common in B-cases (intoxications) than in C-cases (non-intoxications). Note that no
concomitant drugs were found to be significantly more common in C-cases than in B-cases.
Of the concomitant drugs significantly more common in “all B-cases” the painkiller
paracetamol (e.g. Alvedon®) is the only over the counter drug, a drug found in most homes.
The other concomitant drugs require prescriptions and are either sedatives/hypnotics or
analgesics; Dextropropoxyphene: e.g. Dexofen®, deregistered, analgesic; Codeine: e.g.
Citodon®, analgesic; Flunitrazepam: Rohypnol®, deregistered or Flunitrazepam Mylan®,
hypnotic and sedative benzodiazepine; Propiomazine: Propavan®, sedative; Alimemazine:
e.g. Theralen®, drug with unspecific sedative effect. [20]
The concomitant drugs more common in B-cases than in C-cases for the different
index-antidepressant were somewhat repetitive and included the drugs seen in the general
groups listed above.
Table 5: Shows all drugs that were significantly more common in B-cases than in C-cases. Some drugs are marked with “X”, these drugs are listed in SFINX [19] (appendix I), for these specific interactions the clinical significance is unknown or varies.
19
When examining Table 5 only 4 interactions are known and listed in SFINX [19] (appendix I),
these are marked with “X”. The clinical significance of these 4 interactions is unknown or
varies and therefore they should be further studied.
For amitriptyline only the concomitant drug ethanol was more common in B-cases than in
C-cases. Ethanol is known to increase the blood plasma concentration of amitriptyline, by
decreasing the hepatic clearance of amitriptyline [21]. Hence, the combination of
amitriptyline and alcohol (ethanol) increases the risk for intoxication of amitriptyline.
Paracetamol or rather one of the metabolites is known to be toxic by itself [22, 23]. At the
moment there are no known interactions between paracetamol and any of the
index-antidepressants listed in SFINX [19]. For paroxetine, paracetamol was more common
in B-cases than in C-cases. Paracetamol was detected in almost half of the B-cases with
paroxetine, 15 cases with paracetamol and 16 without paracetamol. On the contrary, among
the C-cases only 1 of 34 samples contained paracetamol. This pattern was not seen for any of
the other index-antidepressants. This could, however, be a random finding.
Propiomazine does not have any known drug interactions listed in SFINX [19]. However,
interaction studies with propiomazine are sparse and limited: A PubMed search May 13,
2014 with the search criteria “Propiomazine” and “Interaction” gave 4 hits, none of them
with particular relevance. Propiomazine is more common in B-cases than in C-cases for the
antidepressant drugs; mirtazapine, venlafaxine, citalopram and fluoxetine.
Flunitrazepam was more common in B-cases than in C-cases with the antidepressant drugs
mianserin, mirtazapine, venlafaxine and citalopram. Flunitrazepam is overall one of the most
common drugs found in fatal poisonings [24]. Therefor it is not surprising that flunitrazepam
is overrepresented in the fatal poisonings cases in this study. But still flunitrazepam comes
out as more common in fatal poisonings in combination with some antidepressant drugs.
Concomitant use of tramadol with antidepressant drugs is not recommended, at least not
without initial monitoring [25]. In SFINX codeine is recommended instead of tramadol, as a
painkiller, if venlafaxine or sertraline are also administered to the patient [19]. Codeine was
more common in B-cases than in C-cases for the antidepressant drugs mirtazapine,
venlafaxine and sertraline. Codeine has no known interactions with these
index-antidepressants in SFINX [19]. In the data there are no C-cases with codeine and
sertraline (7 of 46 B-cases and 0 of 124 C-cases) or with codeine and mirtazapine (12 of 67
B-cases and 0 of 162 C-cases).
Venlafaxine in combination with alcohol has been associated with an elevated risk for fatal
poisoning [26]. This cannot be confirmed in this study, because ethanol (alcohol) was not
more common in B-cases than in C-cases.
The painkiller dextropropoxyphene was significantly more common in B-cases than in
C-cases when taken together with a majority of the index-antidepressants (Table 5). It is
20
reasonable to think that dextropropoxyphene could be dangerous by itself and might have
interactions with the index-antidepressants or with the concomitant drugs.
Dextropropoxyphen is known to be one of the most common drugs found in fatal poisonings
[24]. The data studied was sampled between 1992 and 2005. In 2009 a EU-decision was made
to with draw dextropropoxyphene from the market [27, 28]. Due to that the risks, especially
the risk of potentially fatal overdoses, are greater than the benefits of the drug. The findings
in this project strengthen that decision.
A drawback with the study is the uncertainties with regards to the validities of the blood
concentrations of the index-antidepressants. First of all, the doses taken are not known and
most B-cases and C-cases are obvious suicides [2]. Blood concentration for high potentially
toxic doses can be very different from blood concentrations studied in trials of therapeutic
doses. This because differences in pharmacokinetics might occur for drugs in high doses
compared to therapeutic doses [29]. Lastly, blood samples were taken after death so
postmortem drug redistribution cannot be ruled out. This can affect the blood concentration
of substances, for example tricyclic antidepressants [30]. Concentration differences were
therefore not discussed.
8. Conclusions
In conclusion this study shows drugs that were more common in combination with specific
antidepressant drugs in cases of intoxication compared to non-intoxications. These
combinations could have underlying drug-drug interactions causing the intoxication.
Therefore, further studies on the following drug combinations are recommended: Paroxetine
- paracetamol, mirtazapine - flunitrazepam, venlafaxine - flunitrazepam, citalopram -
flunitrazepam, mirtazapine - codeine and sertraline - codeine. Also the 4 known interactions
between: sertraline - zolpidem, venlafaxine - amitriptyline, venlafaxine - nortriptyline and
venlafaxine - citalopram should be studied further. It is important to point out that these
combinations cannot be confirmed to be more dangerous than others. But they should be
considered as interesting for further studies.
This research method probably comes best into place in cases like the one with
propiomazine, which also is the most interesting finding in this paper. Propiomazine is a
poorly studied drug and it is more common as a concomitant drug in B-cases than in C-cases
for 4 antidepressants drugs. Therefore, interaction studies are recommended to be conducted
on the following drug combinations: mirtazapine - propiomazine, venlafaxine -
propiomazine, citalopram - propiomazine and fluoxetine - propiomazine.
9. Acknowledgments
I want to thank my dedicated supervisor Docent
Margareta Reis
for discussing ideas and
an overall great time. Also I want to thank Martin Singull for his statistical support.
21
10. References
1.
Druid, H., P. Holmgren, and P. Lowenhielm, Computer-assisted systems for forensic
pathology and forensic toxicology. J Forensic Sci, 1996. 41(5): p. 830-6.
2.
Reis, M., et al., Reference concentrations of antidepressants. A compilation of postmortem
and therapeutic levels. J Anal Toxicol, 2007. 31(5): p. 254-64.
3.
WHO, International Classification of Diseases, 9th revision (ICD-9). 1979: WHO,
Geneva, Switzerland.
4.
Rang, H.P., M.M. Dale, and S. Elsevier, Rang & Dale's pharmacology. 7th ed. 2012,
Edinburgh ; New York: Elsevier/Churchill Livingstone. xxii, 777 p.
5.
Eyer, F., et al., Risk assessment of severe tricyclic antidepressant overdose. Hum Exp
Toxicol, 2009. 28(8): p. 511-9.
6.
Whyte, I.M., A.H. Dawson, and N.A. Buckley, Relative toxicity of venlafaxine and
selective serotonin reuptake inhibitors in overdose compared to tricyclic antidepressants.
QJM, 2003. 96(5): p. 369-74.
7.
Amrein, R., et al., Interactions of moclobemide with concomitantly administered medication:
evidence from pharmacological and clinical studies. Psychopharmacology (Berl), 1992. 106
Suppl: p. S24-31.
8.
Bleakley, S., Review of the choice and use of antidepressant drugs. Progress in Neurology
and Psychiatry, 2013. 17(6): p. 18-26.
9.
Volpi-Abadie, J., A.M. Kaye, and A.D. Kaye, Serotonin Syndrome. Ochsner J, 2013.
13(4): p. 533-540.
10.
Hollister, L.E., Current antidepressant drugs: their clinical use. Drugs, 1981. 22(2): p.
129-52.
11.
Mathers, C.D. and D. Loncar, Projections of global mortality and burden of disease from
2002 to 2030. PLoS Med, 2006. 3(11): p. e442.
12.
Jones, A.W., et al., Drug poisoning deaths in Sweden show a predominance of ethanol in
mono-intoxications, adverse drug-alcohol interactions and poly-drug use. Forensic Sci Int,
2011. 206(1-3): p. 43-51.
13.
Altman, D.G., Practical statistics for medical research. 1999, Boca Raton, Fla.: Chapman
& Hall/CRC. xii, 611 p.
14.
Viera, A.J., Odds ratios and risk ratios: what's the difference and why does it matter? South
Med J, 2008. 101(7): p. 730-4.
15.
Schmidt, C.O. and T. Kohlmann, When to use the odds ratio or the relative risk? Int J
Public Health, 2008. 53(3): p. 165-7.
16.
Hogue, C.J., D.W. Gaylor, and K.F. Schulz, Estimators of relative risk for case-control
studies. Am J Epidemiol, 1983. 118(3): p. 396-407.
17.
Greenland, S., D.C. Thomas, and H. Morgenstern, The rare-disease assumption revisited.
A critique of "estimators of relative risk for case-control studies". Am J Epidemiol, 1986.
124(6): p. 869-83.
18.
Verhoeven, K.J.F., Simonsen, K. L. and McIntyre, L. M., Implementing false discovery
rate control: increasing your power. OIKOS, 2005. 108: p. 643-647.
19.
SFINX (Swedish, Finnish, Interaction X-referencing). 2014-04-04, :
http://www.janusinfo.se/Beslutsstod/Interaktioner/Interaktioner-Sfinx/
.
20.
FASS – the Swedish Medicines Information Engine. 2014-05-21:
http://www.fass.se
.
21.
Dorian, P., et al., Amitriptyline and ethanol: pharmacokinetic and pharmacodynamic
22
22.
Bender, R.P., et al., N-acetyl-p-benzoquinone imine, the toxic metabolite of acetaminophen, is
a topoisomerase II poison. Biochemistry, 2004. 43(12): p. 3731-9.
23.
Dart, R.C., et al., Acetaminophen poisoning: an evidence-based consensus guideline for
out-of-hospital management. Clin Toxicol (Phila), 2006. 44(1): p. 1-18.
24.
Jonasson, B., U. Jonasson, and T. Saldeen, Among fatal poisonings dextropropoxyphene
predominates in younger people, antidepressants in the middle aged and sedatives in the
elderly. J Forensic Sci, 2000. 45(1): p. 7-10.
25.
Park, S.H., R.C. Wackernah, and G.L. Stimmel, Serotonin syndrome: is it a reason to
avoid the use of tramadol with antidepressants? J Pharm Pract, 2014. 27(1): p. 71-8.
26.
Koski, A., E. Vuori, and I. Ojanpera, Newer antidepressants: evaluation of fatal toxicity
index and interaction with alcohol based on Finnish postmortem data. Int J Legal Med, 2005.
119(6): p. 344-8.
27.
Läkemedelsverket. Dextropropoxifen dras bort från marknaden. 2010-02-15 [cited 2014
05-10].
28.
Agency, E.M. European Medicines Agency recommends withdrawal of
dextropropoxyphene-containing medicines. 2009-06-26 [cited 2014 05-10].
29.
Sue, Y.J. and M. Shannon, Pharmacokinetics of drugs in overdose. Clin Pharmacokinet,
1992. 23(2): p. 93-105.
30.
Yarema, M.C. and C.E. Becker, Key concepts in postmortem drug redistribution. Clin
Toxicol (Phila), 2005. 43(4): p. 235-41.
23
Appendix I – Known drug interactions
Known relevant drug interactions with the index-antidepressants found in SFINX (Swedish,
Finnish, Interaction X-referencing). This database is under constant development and these
interactions could be found April 4, 2014.
Table below shows known drug interactions with the index-antidepressants and the clinical significance meaning. A: The interaction has no clinical significance
B: The interactions clinical significance is unknown and/or varies
C: Clinically important interaction that can be dealt with, for example by dose adjustment D: Clinically meaningful interaction that should be avoided
index-antidepressant
drug interaction with
clinical significance
Amitriptyline
Moclobemide
D
Paroxetine
C
Levomepromazine
C
Fluoxetine
C
Tramadol
C
Carbamazepine
C
Amphetamine
C
Morphine
B
Venlafaxine
B
Fluvoxamine
B
Mirtazapine
A
Citalopram
Moclobemide
D
Fluoxetine
D
Paroxetine
D
Carbamazepine
C
Tramadol
C
Clomipramine
B
Venlafaxine
B
Imipramine
B
Clomipramine
Fluvoxamine
D
Tramadol
D
Moclobemide
D
Codeine
C
Fluoxetine
C
Paroxetine
C
Levomepromazine
C
Venlafaxine
C
Citalopram
B
Morphine
B
Carbamazepine
B
24
Fluoxetine
Desipramine
D
Moclobemide
D
Codeine
D
Venlafaxine
D
Sertraline
D
Tramadol
D
Fluvoxamine
D
Citalopram
D
Paroxetine
D
Amitriptyline
C
Imipramine
C
Carbamazepine
C
Alprazolam
C
Nortriptyline
C
Clomipramine
C
Trimipramine
C
Maprotiline
C
Diazepam
B
Mirtazapine
B
Fluvoxamine
Clomipramine
D
Moclobemide
D
Mirtazapine
D
Fluoxetine
D
Paroxetine
D
Theophylline
D
Caffeine
C
Imipramine
C
Diazepam
C
Alprazolam
C
Tramadol
C
Maprotiline
C
Trimipramine
C
Desipramine
B
Amitriptyline
B
Venlafaxine
B
Imipramine
Carbamazepine
C
Fluoxetine
C
Fluvoxamine
C
Paroxetine
C
Tramadol
C
Venlafaxine
B
25
Maprotiline
Moclobemide
C
Fluvoxamine
C
Fluoxetine
C
Levomepromazine
C
Paroxetine
C
Venlafaxine
B
Mianserin
Moclobemide
D
Carbamazepine
C
Thioridazine
C
Mirtazapine
Fluvoxamine
D
Moclobemide
D
Carbamazepine
C
Venlafaxine
B
Fluoxetine
B
Amitriptyline
A
Moclobemide
Venlafaxine
D
Fluoxetine
D
Clomipramine
D
Fluvoxamine
D
Paroxetine
D
Citalopram
D
Sertraline
D
Tramadol
D
Mianserin
D
Mirtazapine
D
Nortriptyline
D
Amitriptyline
D
Maprotiline
C
Trimipramine
C
Dextropropoxyphene
B
Nortriptyline
Moclobemide
D
Carbamazepine
C
Paroxetine
C
Levomepromazine
C
Fluoxetine
C
Dextropropoxyphene
B
Venlafaxine
B
26
Paroxetine
Venlafaxine
D
Tramadol
D
Desipramine
D
Codeine
D
Moclobemide
D
Fluvoxamine
D
Fluoxetine
D
Sertraline
D
Citalopram
D
Amitriptyline
C
Nortriptyline
C
Imipramine
C
Clomipramine
C
Trimipramine
C
Sertraline
Moclobemide
D
Fluoxetine
D
Paroxetine
D
Carbamazepine
C
Tramadol
C
Desipramine
B
Venlafaxine
B
Zolpidem
B
Trimipramine
Moclobemide
C
Fluoxetine
C
Tramadol
C
Paroxetine
C
Venlafaxine
B
Venlafaxine
Paroxetine
D
Moclobemide
D
Fluoxetine
D
Clomipramine
C
Desipramine
B
Carbamazepine
B
Imipramine
B
Mirtazapine
B
Trimipramine
B
Nortriptyline
B
Amitriptyline
B
Citalopram
B
Sertraline
B
Fluvoxamine
B
Maprotiline
B
27
Appendix II – Relative risk
Relative risk for the prevalence of concomitant drugs in B-cases compared to C-cases.
When studying the following relative risk plots remember that significant findings have to
meet the two criteria that are listed below:
1. The confidence interval of the relative risk´s, showed as blue horizontal lines, has to
be separated from 1. This means that the confidence interval should not overlap the
blue vertical line in the plot.
2. The p-value (shown in the side of the plot for each drug) has to be lower than the
value from the Benjamini-Hochberg procedure which can be seen in the header of the
plot. To make it easier drugs that meet this criterion will be below a horizontal black
line in the plot. If no black line is seen no p-values were significant after the
Benjamini-Hochberg correction.
Drugs which have 95% confidence intervals not overlapping are significantly more or less
common in B-cases or C-cases.
In the right-hand side of the plot the name and values for the drugs are listed (in the same
order as the showed confidence intervals); a, c, b and d are the number of cases
28
Figure A: The drugs dextropropoxyphene, ethanol, codeine, flunitrazepam, paracetamol, propiomazine, alimemazine are signifcantly more common in B-cases (intoxications) than in C-cases (nonintoxications).
29
30
Figure C: Dextropropoxyphene, ethanol, levomepromazine, propiomazine, flunitrazepam, paracetamol, diazepam and alimemazine are significantly more common in B-cases than C-cases.