SEARCH

SEARCH BY CITATION

Keywords:

  • serotonin toxicity;
  • prescriptions;
  • drug utilization;
  • drug interactions;
  • aged;
  • veterans

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. REFERENCES
  8. Supporting Information

WHAT IS ALREADY KNOWN ABOUT THIS SUBJECT

• Serotonin toxicity is greatly increased by the co-prescribing of serotonergic agents.

• Serotonin toxicity varies greatly in severity from mild to potentially life-threatening, however even mild cases can impair quality of life, especially for older people, by causing agitation and sleep disturbance. Combinations including MAOI are most likely to cause severe toxicity.

• Few studies have used pharmaceutical claims data to quantify the use of serotonergic medicines in combination.

WHAT THIS STUDY ADDS

• In a cohort of elderly Australian veterans and dependants, we found widespread use of serotonergic medicines: 115 969 (42%) of the study population (n = 273 228) were dispensed at least one of these medicines between July 2000 and June 2004.

• Approximately 8% (20 658 individuals) experienced at least one episode of potential concomitant use of serotonergic medicine combinations.

• Potentially life-threatening combinations involving MAOIs were of concern: 1811 (0.7%) individuals had at least one episode with such combinations.

• This study demonstrates the utility and benefits of pharmaceutical claims data to provide insights into real-world prescribing.

AIMS

We examined potential risk of serotonin toxicity in Australian veterans by quantifying the concomitant use of serotonergic medicine combinations from claims data collected by the Department of Veterans’ Affairs (DVA).

METHODS

This was a retrospective cohort study of 273 228 Australian veterans, war widows, widowers and dependants aged ≥55 years and holding full treatment entitlement for the period July 2000 to June 2004 or until death. The main outcome measure was potential concomitant use, estimated as the number of cohort members with an overlap in days of supply for serotonergic medicine combinations over the 4 year period for all medicine combinations and potentially life threatening combinations.

RESULTS

From July 2000 to June 2004, 115 969 (42%) cohort members were dispensed at least one serotonergic medicine. 20 658 (8%) had at least one episode of potential concomitant use. We identified 1811 (0.7%) cohort members with at least one overlapping period of potentially life-threatening serotonergic medicine combinations, 937 of whom had the combinations dispensed within the recommended washout period. Three hundred and seventeen of these individuals were dispensed potentially life-threatening medicine combinations on the same day. The most common combinations were moclobemide with a selective serotonin reuptake inhibitor or tramadol.

CONCLUSIONS

The individuals potentially at risk of mild to moderate serotonin toxicity were considerable and potentially life threatening combinations were not infrequent. While we were unable to determine how many individuals experienced serotonin toxicity this study indicates, for the first time, the potential size of the problem in a subgroup of elderly Australians. Clinicians and patients need to be vigilant regarding inadvertent concomitant use, especially that of moclobemide with a selective serotonin reuptake inhibitor or tramadol.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. REFERENCES
  8. Supporting Information

Certain commonly prescribed medicines, used alone or in combination, carry a risk of serotonin toxicity [1, 2]. These medicines are often co-prescribed to treat pain, depression and anxiety, conditions which frequently coexist, and where more than one doctor is likely to be involved in treatment [3, 4]. In Australia, the use of serotonergic medicines is considerable. For example, in 2005 more than 7 million prescriptions for selective serotonin reuptake inhibitors (SSRIs) and 2 million tramadol prescriptions were dispensed from community pharmacies [5].

Serotonin syndrome is a predictable consequence of serotonin toxicity arising from increased concentrations of serotonin at neurological receptor sites [1, 3]. Serotonin toxicity varies greatly in severity from mild to potentially life-threatening [6–8]. However, even mild cases can impair quality of life, especially for older people, by causing agitation and sleep disturbance. Unfortunately such unrecognized cases may be managed by prescribing patients yet another medicine, further increasing the potential for an adverse event.

In principle, moderate to severe serotonin toxicity is largely avoidable [1, 9]. It is most often found in patients taking two or more medicines at therapeutic doses that increase CNS serotonergic activity by different mechanisms [2, 8, 9]. Although many drugs are stated to be serotonergic, there is a great deal of variation even within drug classes and some combinations which carry warnings probably have a very low likelihood of causing problems. For example, within the tricyclic antidepressant class serotonin re-uptake inhibition varies considerably, imipramine and clomipramine being by far the most potent and the only agents clearly linked to life-threatening interactions [10]. The most serious cases of serotonin toxicity following drug interactions have involved non-selective MAOIs or selective MAO-A inhibitors combined with another potent serotonergic drug with a different mechanism [10].

Currently in Australia, the primary source of information about the occurrence of adverse events including serotonin toxicity is the Therapeutic Goods Administration (TGA) voluntary reporting system. Clearly this is not a comprehensive list of cases and the database probably reflects the more serious end of the spectrum. However, risk quantification requires measurement of incidence of exposure, as well as strength of association between exposure and outcome. Such information is not available from adverse event reporting databases. Greater access to existing health administrative data sources to explore potential risks associated with medicine use would complement the existing pharmacovigilance strategies [11]. A recent study of Australian veterans [12] illustrated the potential of dispensing data in quantifying this risk over a 6 month period. The study examined the prevalence of duplicate antidepressant prescribing and potentially avoidable drug interactions such as anticholinergic agents with tricyclic antidepressants and tramadol with antidepressants. Our study advances this research by examining a broader range of serotoninergic drugs and drug combinations, over a longer observation period, and using different algorithms to quantify the extent of the problem.

Prescription data for dispensed medicines are held for all Australians who claim services under the Repatriation Pharmaceutical Benefits Schedule (RPBS). These data have recently become available for research and provide a unique opportunity for evaluation of medicines use, safety and outcomes [12, 13]. In this paper we examined exposure to serotonergic medicines, either as sole agents or in combination, for a cohort of elderly Australians, using pharmaceutical claims data collected by the Department of Veterans’ Affairs (DVA).

Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. REFERENCES
  8. Supporting Information

Ethics

This study was approved by the Human Research Ethics Committees of the Department of Veterans’ Affairs and the University of New South Wales.

Data holdings

We extracted data from the DVA client and pharmaceutical claims databases. The client database contains information on entitled veterans including: unique card holder identifier, dates of birth and death, sex, entitlement level and the start and end date of each continuous period at a specific entitlement level for each client. The pharmaceutical claims files contain data on dispensed pharmaceutical items including: unique card holder identifier, entitlement at time each item was dispensed, sex, date of birth, pharmaceutical item details (item code, name and strength, Anatomical Therapeutic Chemical (ATC) code, date of supply, packs supplied and number of repeats).

Medicines of interest

Based on the literature [1, 2, 6–10, 14–17] we identified medicines associated with serotonin toxicity (Table S1). Drug classes included tricyclic antidepressants (non-selective monoamine reuptake inhibitors, hereafter referred to as TCAs), SSRIs, non-selective monoamine oxidase inhibitors (MAOIs), monoamine oxidase A inhibitors (moclobemide), other antidepressants (ADs), and 5-hydroxytryptamine 1 agonists (5-HT1 agonists). Tramadol, fentanyl, pethidine, selegiline, lithium, linezolid, pentazocine, dexamethorphan and sibutramine were also examined.

Cohort selection

For the period July 2000 to June 2004 we identified a cohort of veterans, war widows, widowers and dependants holding full treatment entitlement (Gold Card Holders; GCH) aged ≥55 years and enrolled for the duration of the study period or until death. These veterans are entitled to claim health care, related services and pharmaceutical items for all health needs, whether related to military service or not. Therefore, data are available for all prescribed medicines dispensed to GCHs under the Repatriation and Pharmaceutical Benefits Schemes (R/PBS).

Period prevalence of serotonergic medicine use

We identified the number of cohort members dispensed serotonergic medicines of interest at least once during the four-year period, overall and according to specific medicine categories (as defined in Table S1). We also calculated the number of serotonergic medicine categories cohort members were dispensed over the period.

Concomitant use

We developed a model to determine potential concomitant use of medicine combinations of interest, based on the following assumptions: medicine use commenced on the date of supply; and all medicines were consumed at a rate of one Defined Daily Dose (DDD) per day or one tablet per day (if tablet strength was greater than one DDD). The DDD is the international unit of drug utilization and is the assumed daily dose of a medicine when used for its main indication by adults [18]. DDDs for the medicines of interest are listed in Table S1. We examined two possible scenarios regarding the dispensing of more than one prescription for the same medicine dispensed on the same day or prior to the expected completion of a previous prescription. In the first scenario, we assumed medicines were taken in parallel (for example, the medicine was consumed at twice the rate, or double the dose). In the second, we applied the assumption that medicines were consumed in sequence (i.e. the medicine was consumed at a rate of one DDD or one tablet per day). Our calculations show concomitant use estimates were similar for the two scenarios. Therefore, we only report estimates derived from the parallel consumption model.

Based on this model we calculated the use of serotonergic medicine combinations in the following ways:

  • 1
    The number of cohort members dispensed at least two serotonergic medicine categories (Table S1) on the same day or with any potential overlap in days supply.
  • 2
    The number of cohort members with any overlap in days supply for specific medicine combinations and potentially life-threatening combinations (namely MAO-inhibitors and other drugs which strongly inhibit the reuptake of serotonin [10]).
  • 3
    The number of cohort members with potentially life-threatening combinations dispensed on the same day or within the recommended wash-out period for that combination [19], (i.e. for which no assumption for drug interaction potential needed to be made other than current use at the time of dispensing).

All analyses were performed using SAS software, version 9 (SAS Institute Inc, North Carolina, USA).

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. REFERENCES
  8. Supporting Information

Demographic details of our study cohort are summarized in Table 1. Not unexpectedly, there were more males than females and more than 80% were aged 75 years and over.

Table 1.  Demographic characteristics of study cohort* (n = 273 228)
 n%
  • *

    study cohort = veterans, war widows, widowers and dependants holding full treatment entitlement (Gold Card Holders; GCH) aged ≥55 years and enrolled continuously for the duration of the study period or until death.

SexMale165 32460.5
Female107 90439.5
Age group (at July 2000)55–596 0462.2
60–645 0311.8
65–699 5703.5
70–7431 35511.5
75–79113 48041.5
80–8472 03926.4
85+35 70713.1

Period prevalence of serotonergic medicine use

Over the four-year study period 115 969 (42.4%) cohort members received at least one serotonergic medicine and the proportion dispensed SSRIs, TCAs, ‘other antidepressants’ and moclobemide was 18%, 15%, 6% and 2% respectively. Further, 18% received at least one tramadol prescription, with fewer than 0.5% of cohort members dispensed non-selective MAOIs, selegiline, pethidine, 5-HT1 agonists, lithium and fentanyl. No cohort members were dispensed linezolid, pentazocine, dextromethorphan, sibutramine or fenfluramine (Table 2).

Table 2.  Period prevalence of serotonergic medicine use in study cohort (n = 273 228), July 2000 to June 2004
 n%
  • *

    Subcategorization of these medicines has occurred as they are more clearly implicated in serotonin syndrome.

Monoamine oxidase inhibitors (MAOIs), non-selective1410.1
Moclobemide5 7102.1
Selective serotonin reuptake inhibitors (SSRIs)48 14917.6
Other antidepressants (Other ADs)16 3896.0
 Venlafaxine*8 0412.9
 Others (mianserin, nefazodone, mirtazapine)9 3283.4
Pethidine9120.3
Tramadol49 91218.3
Fentanyl4 2101.5
Lithium9210.3
Selegiline3980.1
Tricyclic antidepressants (TCAs)40 59514.9
 Imipramine, clomipramine*5 8522.1
 Others (desipramine, amitriptyline, nortriptyline, doxepin, dothiepin)35 87313.1
Selective serotonin (5HT1) agonists6430.2
Linezolid00.0
Pentazocine00.0
Dextromethorphan00.0
Sibutramine00.0
Fenfluramine00.0
Any serotonergic medicine115 96942.4

We identified 76 621 (28%) cohort members dispensed medicines from only one serotonergic medicine category over the four-year period and 10% (28 179), 3% (8343) and 1% (2826) from two, three and four or more categories. However, this dispensing did not necessarily reflect concomitant use as the medicines could have been dispensed at any time during the period.

Concomitant use

Eight percent (20 658) of cohort members were dispensed medicines from different medicine categories with some temporal overlap in use. Five percent (12 222), 2% (5919) and 1% (2517) of cohort members were dispensed two, three and four or more medicines types with at least one period of overlap in the four-year period. Furthermore, of those who had at least one episode of temporal overlap from different medicine categories, 50% had at least one episode of same day dispensing. Twenty percent (5497) of cohort members supplied medicines from two categories and 38% of those supplied medicines from three categories had at least one episode of same day dispensing (Table 3).

Table 3.  Number of cohort members with temporal overlap and same day dispensing by number of serotonergic medicine categories* dispensed
Number of medicine categoriesnSame day dispensingAny temporal overlap
n%n%
  • *

    Medicine categories are detailed in Table S1.

0157 259
176 621
228 1795 49719.512 22243.4
38 3433 11837.45 91970.9
42 1791 12351.51 89587.0
553433562.751195.7
6+1138575.211198.2
All273 22810 1583.720 6587.6

Medicines most commonly used concomitantly included SSRIs (n = 14 326), tramadol (11 777) and TCAs other than imipramine and clomipramine (9173). We identified 1811 cohort members with at least one period of overlap of potentially life threatening serotonergic medicine combinations. Specifically, we identified those with at least one episode of overlap between moclobemide and SSRIs (n = 1069), tramadol (n = 561), venlafaxine (n = 298), imipramine/clomipramine (n = 71) and pethidine (n = 4). In addition, we found cohort members with at least one period of overlap between MAOIs and SSRIs (n = 18), tramadol (n = 17), venlafaxine (n = 11) and imipramine or clomipramine (n = 1) (Table 4).

Table 4.  Number of cohort members potentially using medicine combinations concomitantly* Thumbnail image of

We examined the temporal relationship between the dispensing of potentially life-threatening medicine combinations to find 317, of the 937 cohort members dispensed potentially very serious interactions up to ten days apart, were dispensed these medicines on the same day (Table 5). An additional 20% (n = 205) were dispensed medicines before the recommended washout period had elapsed from the previous dispensing.

Table 5.  Number of cohort members dispensed potentially life-threatening medicine combinations within recommended washout periods Thumbnail image of

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. REFERENCES
  8. Supporting Information

This study illustrates how pharmaceutical claims data can help assess the risk of serotonin toxicity in a defined population. While we were unable to determine how many individuals actually experienced serotonin toxicity the study indicates, for the first time, the potential size of the problem in the Australian veteran population. Quantification of the number of individuals who may be at risk will help guide regulatory and educational bodies in alerting prescribers, patients, carers, and pharmacists to be vigilant.

During the four-year study, over 40% of the cohort received at least one serotonergic medicine. Concomitant use of serotonergic medicines (as indicated by one or more overlapping periods of prescription supply) was frequent. The numbers of individuals at risk of mild to moderate serotonin toxicity were therefore considerable. The most common combinations were: tramadol and a SSRI; tramadol and a TCA; a TCA and a SSRI; a SSRI and a medicine from the other antidepressants class; and a combination of two SSRIs. Potentially lethal combinations were also observed, including moclobemide/MAOI with a SSRI or venlafaxine or tramadol. Many of these were co-dispensed without any possible time for the recommended wash-out period to have occurred (Table 5). That 317 patients received such combinations on the same day is of concern, suggesting that the interaction risk was overlooked by both prescribers and pharmacists.

Overall, 20 658 people had overlapping combinations of serotonergic drugs. It is unlikely all were due to oversights. Many combinations are relatively safe and experienced prescribers may have decided the benefits outweighed the risks of co-prescription. For example, two prescriptions for medicines from different antidepressant drug classes [12, 20] might commonly occur transiently while switching medications [2, 7, 9, 21].

Some combinations are more likely than others to occur inadvertently. One in five individuals in this elderly cohort was dispensed at least one SSRI or tramadol prescription at least once over the four-year period. Combinations involving overlap of antidepressants with tramadol were consequently very common and were unlikely to represent planned switching of prescribed medication. In the four years following its marketing in Australia in late 1998, tramadol gave rise to 20 reports of serotonin syndrome, in 16 of which the patient was taking potentially interacting medicines such as antidepressants, sibutramine and St John's Wort [21].

Tramadol is readily available and recommended in pain management guidelines [22]. Patients and doctors need to be aware that tramadol can contribute to the risk of serotonin toxicity and be watchful for symptoms and signs [21–24]. Based on our findings, particular attention should be paid to the quite high possibility that older patients may be taking tramadol (or fentanyl) in combination with a serotonergic antidepressant. The risk-benefit of such combinations is very likely to be much worse in terms of both safety and efficacy than when these drugs are prescribed on their own.

Medicines such as linezolid [9, 16, 17], lithium, dextromethorphan, selegiline, pethidine and 5-HT1 agonists for use in migraine have been implicated as placing patients at high risk of serotonin toxicity [1, 7, 9], but were dispensed rarely or not at all in this cohort.

A limitation of our study is that we cannot know if cohort members actually took the medicines dispensed, and if so, for how long the medicine was taken and the actual daily doses. Concomitant medicine use can only be inferred by the overlap in supply of medicines as estimated from the recorded dispensing intervals. Such methodological limitations are common to pharmacoepidemiological studies using dispensing and other healthcare databases to determine medicine exposure. However, even using the most conservative assumptions about duration of use (current at time of dispensing), there were more than 20 000 people with overlap in duration of two serotonergic drugs, and 1800 people who had overlaps of the most potentially serious combinations, with more than 300 people receiving such combinations on the same day.

Given the limited data on the incidence of serotonin toxicity [1, 7] we were not able to compare the observed frequency of potential concomitant use with the observed incidence of serotonin toxicity. With respect to the generalizability of our findings, our study population is older than the general Australian population as we selected those 55 years and over. However, in 2003 DVA beneficiaries represented approximately 10% of the total number of Australians aged 65 years and above, and 25% of Australians aged 80 years and over [25]. As such these findings may well reflect an important clinical issue for all elderly Australians. Clearly, as polypharmacy increases with age this is the population at greatest risk.

Our study highlights the utility of person level data in informing pharmacovigilance. The Department of Veterans’ Affairs database is unique in Australia in that it can, subject to ethical approval, provide comprehensive person level data with the appropriate safeguards to protect the identity of patients and their treating doctors. Improved pharmacovigilance in Australia demands a national effort to overcome the barriers to undertaking these and other pharmacoepidemiological studies using the Medicare Australia databases [11, 26, 27]. Such studies provide an estimate of the potential size of a clinically relevant problem and therefore inform educational strategies to reduce numbers at risk. Given limited funding for dedicated pharmacovigilance studies, it is imperative that existing data repositories are used effectively to help ensure safe use of medicines.

Professor David Henry advised on the analytical approach and Professor Richard Day and Dr Susan D. Whicker reviewed an earlier version of the manuscript. Valerie Rendle and Louise Snowden assisted the literature review; Dr Margaret Ward Curran provided the summary of serotonin syndrome case reports from the Adverse Drug Reactions Database. The study was assisted by a research grant from the Commonwealth Department of Veterans’ Affairs which supplied claims data and project funding.

REFERENCES

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. REFERENCES
  8. Supporting Information
  • 1
    Boyer EW, Shannon M. The serotonin syndrome. N Engl J Med 2005; 352: 111220.
  • 2
    Mitchell PB. Drug interactions of clinical significance with selective serotonin reuptake inhibitors. Drug Saf 1997; 17: 390406.
  • 3
    Ener RA, Meglathery SB, Van Decker WA, Gallagher RM. Serotonin syndrome and other serotonergic disorders. Pain Med 2003; 4: 6374.
  • 4
    Nieuwstraten C, Labiris NR, Holbrook A. Systematic overview of drug interactions with antidepressant medications. Can J Psychiatry 2006; 51: 30016.
  • 5
    Commonwealth of Australia. Australian Statistics on Medicines 2004–05. AIHW Catalogue Number 47. Canberra: AIHW, 2007.
  • 6
    ADRAC. Serotonin syndrome. Aust Adv Drug React Bull 2004; 23: 34.
  • 7
    Hall M, Buckley N. Serotonin syndrome. Aust Prescr 2003; 26: 623.
  • 8
    Isbister GK, Buckley NA, Whyte IM. Serotonin toxicity: a practical approach to diagnosis and treatment. Med J Aust 2007; 187: 3615.
  • 9
    Chu J, Wang RY, Hill NS. Update in clinical toxicology. Am J Respir Crit Care Med 2002; 166: 915.
  • 10
    Gillman PK. A review of serotonin toxicity data: implications for the mechanisms of antidepressant drug action. Biol Psychiatry 2006; 59: 104651.
  • 11
    Kelman CW, Pearson S-A, Day RO, Holman CDAJ, Kliewer EV, Henry DA. Evaluating medicines: let's use all the evidence. Med J Aust 2007; 186: 24952.
  • 12
    Roughead EE, McDermott B, Gilbert AL. Antidepressants: prevalence of duplicate therapy and avoidable drug interactions in Australian veterans. Aust N Z J Psychiatry 2007; 41: 36670.
  • 13
    Pearson SA, Ringland C, Kelman C, Mant A, Lowinger J, Stark H, Nichol G, Day R, Henry D. Patterns of analgesic and anti-inflammatory medicine use by Australian veterans. Intern Med J 2007; 37: 798805.
  • 14
    ADRAC. Convulsions and blood dyscrasias with mirtazapine. Aust Adv Drug React Bull 2003; 22: 189.
  • 15
    ADRAC. Sibutramine – four years experience. Aust Adv Drug React Bull 2006; 25: 11.
  • 16
    Clark DB, Andrus MR, Byrd DC. Drug interactions between linezolid and selective serotonin reuptake inhibitors: case report involving sertraline and review of the literature. Pharmacotherapy 2006; 26: 26976.
  • 17
    Taylor JJ, Wilson JW, Estes LL. Linezolid and serotonergic drug interactions: a retrospective survey. Clin Infect Dis 2006; 43: 1807.
  • 18
    Norwegian Institute of Public Health. WHO Collaborating Centre for Drug Statistics Methodology. Oslo: WHO, 2006.
  • 19
    Rossi S (editor). Australian Medicines Handbook. Adelaide: Australian Medicines Handbook Pty Ltd, 2007.
  • 20
    McManus P, Mant A, Mitchell P, Birkett D, Dudley J. Co-prescribing of SSRIs and TCAs in Australia: how often does it occur and who is doing it? Br J Clin Pharmacol 2001; 51: 938.
  • 21
    ADRAC. Tramadol – four years experience. Aust Adv Drug React Bull 2003; 22: 12.
  • 22
    Australian and New Zealand College of Anaesthetists and Faculty of Pain Medicine. Acute Pain Management: Scientific Evidence, Second Edition. Melbourne: Australian and New Zealand College of Anaesthetists, 2005.
  • 23
    ADRAC. Tramadol and serotinin syndrome. Aust Adv Drug React Bull 2001; 20: 267.
  • 24
    Kaye K. Trouble with tramadol. Aust Prescr 2004; 27: 267.
  • 25
    Australian Bureau of Statistics. Australian Demographic Statistics June Quarter 2004. Canberra: ABS, 2005.
  • 26
    Kelman CW, Bass AJ, Holman CDJ. Research use of linked health data – a best practice protocol. Aust N Z J Public Health 2002; 26: 2515.
  • 27
    Stanley FJ, Meslin EM. Australia needs a better system for health care evaluation. Med J Aust 2007; 186: 2201.

Supporting Information

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. REFERENCES
  8. Supporting Information

Table S1 Medicines associated with serotonin toxicity

This material is available as part of the online article

Please note: Blackwell Publishing is not responsible for the content or functionality of any supplementary materials supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article.

FilenameFormatSizeDescription
BCP_3253_sm_Table_S1.doc38KSupporting info item

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.