Advances in Psychiatric Treatment (2007), vol. 13, 336–346  doi: 10.1192/apt.bp.106.003194
`
`Why don’t patients take their medicine?
`Reasons and solutions in psychiatry†
`Alex J. Mitchell & Thomas Selmes
`
`Abstract  Over the course of a year, about three­quarters of patients prescribed psychotropic medication will 
`discontinue, often coming to the decision themselves and without informing a health professional. 
`Costs associated with unplanned discontinuation may be substantial if left uncorrected. Partial non­
`adherence (much more common than full discontinuation) can also be detrimental, although some 
`patients rationally adjust their medication regimen without ill­effect. This article reviews the literature 
`on non­adherence, whether intentional or not, and discusses patients’ reasons for failure to concord 
`with medical advice, and predictors of and solutions to the problem of non­adherence. 
`
`This is the first of two articles by Mitchell & Selmes in APT on patient
`engagement and retention in treatment. The second, which addresses
`missed appointments, will appear in the next issue of the journal
`(Mitchell & Selmes, 2007).
`
`The degree to which an individual follows medical 
`advice is a major concern in every medical specialty 
`(Osterberg & Blaschke, 2005). Much attention has 
`focused on methods to persuade patients to adhere 
`to recommendations, without sufficient acknowl­
`edgement  that  avoidance  of  sometimes  complex, 
`costly  and  unpleasant  regimens  may  be  entirely 
`rational  (Mitchell,  2007a).  Equally  overlooked  is 
`the influence of communication between patients 
`and  healthcare  professionals.  Put  simply,  if  no 
`clear agreement is formed with the patient at the 
`onset of treatment then it should be of no surprise 
`if concordance turns out to be less than ideal. In one 
`study the chance of premature discontinuation was 
`found to be less than half in patients who recalled 
`being told to take the medication for at least 6 months 
`compared  with  those  not  given  this  information 
`(Bull et al, 2002b). This task is made more difficult 
`when patients lack insight into their condition (see 
`below).
`Given  the  necessity  of  therapeutic  agreement, 
`the term compliance has given way to adherence 
`and  concordance  (Box  1)  (Haynes et al,  2002).  In 
`considering  the  nosology  of  concordance  and 
`adherence a useful distinction is between individuals 
`
`† For a commentary on this article see pp. 347–349, this issue.
`
`who do not start a medication (similarly those who 
`do  not  attend  their  first  appointment)  and  those 
`who  start  the  course  but  either  take  medication 
`incompletely (partial compliance or adherence) or 
`discontinue prematurely against medical advice (Fig. 
`1). There has been little research on reluctance to start 
`medication which equates to treatment refusal or 
`overt non­adherence. Kasper et al (1997) found in a 
`group of 348 newly admitted psychiatric in­patients 
`that 12.9% refused treatment but that 90% of these 
`ended their refusal within 4 days. Synthesising data 
`on adherence behaviour is difficult because of the 
`wide range of assessment methods (for review see 
`Velligan et al, 2006). Few studies have examined what 
`
`Box 1 Working definitions about health
`behaviour
`
`Adherence (compliance)  is  the  extent  to  which 
`an individual changes their health behaviour 
`to coincide with medical advice
`Concordance is  the  degree  to  which  clinical 
`advice and health behaviour agrees
`Therapeutic alliance  is  an  agreement  between 
`patients  and  health  professionals  to  work 
`together
`Therapeutic disagreement is a divergence in the 
`views  of  patients  and  doctors  on  the  subject 
`of treatment
`
`Alex Mitchell is Consultant and Honorary Senior Lecturer in Liaison Psychiatry at University Hospitals Leicester (Department of Liaison 
`Psychiatry, Brandon Unit, Leicester General Hospital, Gwendolen Road, Leicester LE5 4PW, UK. Email: alex.mitchell@leicspart.nhs.uk) 
`and author of the prize­winning book Neuropsychiatry and Behavioural Neurology Explained: : Diseases, Diagnosis, and Management (2003). 
`Thomas Selmes is a senior house officer on the York psychiatry scheme and was previously a medical student at Leicester University.
`
`336
`
`

`
`Why don’t patients take their medicine?
`
`Course  
`of medication started
`
`No
`
`Refusal of primary 
`(initial) treatment
`
`Yes
`
`Overt non­adherence
`
`Medication stopped
`
`Medication interrupted
`
`Medication adjusted
`
`Discontinuation
`
`Full non­adherence
`
`Conversion to discontinuation
`
`Missed doses
`
`Extra doses
`
`Partial non­adherence
`
`Excess adherence
`
`Patient wished to stop 
`taking medication?
`
`Yes
`
`No
`
`Patient wished  
`to alter dose?
`
`Yes
`
`No
`
`Intentional
`
`Unintentional
`
`Intentional
`
`Unintentional
`
`External
`
`Internal
`
`Barrier
`
`Lapse or slip
`
`External
`
`Internal
`
`Barrier
`
`Lapse or slip
`
`Disclosed to clinician?
`
`Yes
`
`No
`
`Disclosed to clinician?
`
`Yes
`
`No
`
`Refusal of secondary 
`treatment
`
`Self­discontinuation
`
`Collaborative  
`self­medication
`
`Autonomous  
`self­medication
`
`Overt discontinuation
`
`Covert discontinuation
`
`Overt partial adherence Covert partial adherence
`
`Fig. 1 Nosology of adherence behaviour.
`
`advice patients actually receive and thus it is often 
`impossible to test whether patient behaviour is truly 
`at odds with what was agreed.
`It  is  useful  to  remember  than  many  predictors 
`are generic and applicable to all patients (Box 2). 
`Three important predictors are complexity of the 
`regimen, duration of the course and frequency of 
`follow­up contact. Regular contact reduces rates of 
`both missed medication and missed appointments 
`(Rittmannsberger et al, 2004; this will be discussed 
`further in Mitchell & Selmes, 2007). The complexity 
`of a treatment regimen (number of tablets, number 
`of medications and drug interval) adversely affects 
`adherence. Equally, regimens that require disruption 
`to lifestyle, or special techniques or arrangements 
`are less welcome by patients. A systematic review 
`found that the number of doses prescribed per day 
`was inversely related to adherence (Claxton et al, 
`2001). In elderly patients acceptable adherence is 
`common  in  those  taking  only  one  medicine  but 
`rapidly falls in those taking four or more. Regarding 
`the  length  of  the  prescribed  course  of  treatment, 
`most imagine it is relatively simple to follow a 5­
`day antibiotic course but in reality only two­thirds 
`
`manage to do so successfully. Ask patients to follow 
`a 7­day course, with antibiotics four times a day, 
`and ideal concordance is achieved by less than 40%. 
`In psychiatry, we expect patients to agree to much 
`longer courses of medication (occasionally for an 
`indefinite period if risks are judged sufficiently high) 
`(Mitchell, 2006a). 
`In this review, we will examine the predictors of 
`both partial adherence and discontinuation in those 
`taking psychotropic medication, in an attempt to 
`understand why some patients have difficulty fol­
`lowing the advice of mental health professionals and 
`why adherence rates appear to be lower than in other 
`specialties (see also Cramer & Rosenheck, 1998).
`
`Costs of less-than-ideal adherence
`
`There is considerable evidence that premature medi­
`cation discontinuation is costly (Sullivan et al, 1995). 
`Undisclosed (covert) non­adherence appears to be 
`particularly hazardous (Task Force for Compliance, 
`1994).  Kang  et al  (2005)  examined  antipsychotic 
`adherence  of  individuals  prior  to  admission  to 
`
`Advances in Psychiatric Treatment (2007), vol. 13. http://apt.rcpsych.org/
`
`337
`
`

`
`Box 2 Predictors of treatment concordance
`problems
`
`General
`Duration and complexity of regimen
`Lack of informal support
`Patient (intentional)
`Concerns about the side­effects 
`Few perceived benefits
`Stigma of taking medication
`Adjustment to suit daily routine
`Concerns about cost
`Concerns about availability
`Concerns about dependency
`Patient (non-intentional)
`Slips and lapses
`External distractors
`Misunderstanding instructions
`Clinician
`Poor doctor–patient relationship
`Poor empathy
`Poor explanation/communication
`Inadequate follow­up
`Illness
`Severe illness
`Depression or distress
`Psychosis
`Cognitive impairment
`
`• • • • • • • • • • • • • • • • • • • •
`
`Specific adverse effects
`
`Weight gain due to medication has been linked with 
`non­adherence and subjective distress (Fakhoury et
`al, 2001). Obese individuals are more than twice as 
`likely as those with a normal body mass index to 
`miss their medication (Weiden et al, 2004a). Fakhoury 
`(1999) found that more than 70% of patients described 
`weight  gain  due  to  antipsychotics  as  extremely 
`distressing, which was higher than that for any other 
`side­effect.
`Sexual dysfunction is a significant source of distress 
`and may be linked to poor adherence. Olfson et al 
`(2005) studied sexual dysfunction in 139 out­patients 
`with DSM–IV schizophrenia who were receiving an 
`antipsychotic but no other medications associated 
`with sexual side­effects. Sexual dysfunction occur­
`red in 45.3% of the group and was associated with 
`significantly  lower  ratings  on  global  quality  of 
`life. Rosenberg et al (2003) examined the effects of 
`sexual side­effects on adherence. They found that 
`62.5% of men and 38.5% of women felt that their 
`psychiatric medications were causing sexual side­
`effects; 41.7% of men and 15.4% of women admitted 
`
`Mitchell & Selmes
`
`hospital with schizophrenia: 37.1% had been totally 
`non­adherent during the 6 months prior to admission. 
`Although the majority had missed medication on 
`fewer  than  20%  of  days,  many  relapsed  anyway. 
`Reduced  adherence  weakens  treatment  benefits, 
`especially when no alternatives are explored (Irvine
`et al, 1999). 
`Fewer studies in psychiatry have looked at the 
`complications of partial concordance. Three of the 
`most influential of these have been conducted in 
`the USA. In a sample of 7864 Medicaid patients tak­
`ing atypical antipsychotics, those who achieved less 
`than 80% of ideal concordance were about 50% more 
`likely than fully concordant patients to have been 
`hospitalised (Eaddy et al, 2005). Jeste’s group (Gilmer
`et al, 2004) found a hierarchy of risk of admission, 
`from those who were non­adherent, those who took 
`medication to excess and finally those who were 
`partially adherent. In the largest study, Valenstein et
`al (2002) showed that those with less than 80% con­
`cordance were 2.4 times more likely to be admitted 
`than patients with good concordance. 
`
`General predictors of missed
`medication
`
`A number of authors have reviewed factors predict­
`ing non­adherence, but only a handful have con­
`sidered predictors in psychiatric populations (Fenton 
`et al, 1997). Predictors may be usefully divided into 
`patient factors, clinician factors and illness factors 
`(Box 2).
`
`Patient factors
`Intentional non­adherence
`
`Distinguishing intentional non­adherence (missing 
`or  altering  doses  to  suit  one’s  needs)  from  un­
`intentional non­adherence (e.g. forgetting to take 
`medication) is a relatively recent development. In 
`fact, in long­term studies understandable reasons for 
`discontinuation are more common than irrational 
`reasons. Intentional non­adherence is predicted by 
`the balance of an individual’s reasons for and against 
`taking medication, as suggested by utility theory. 
`Intentional non­adherence is a common reason not 
`to start a course of medication, but it may be less 
`common than accidental non­adherence in relation 
`to  missing  individual  doses  (Lowry et al,  2005). 
`Predictors  of  intentional  non­adherence  include 
`less severe disease (and feeling well), the desire to 
`manage independently of the medical profession 
`(self­efficacy),  disagreement  with  or  low  trust 
`in clinicians, and receipt of low levels of medical 
`information (Piette et al, 2005). 
`
`338
`
`Advances in Psychiatric Treatment (2007), vol. 13. http://apt.rcpsych.org/
`
`

`
`that  they  had  stopped  their  medications  at  some 
`point  during  treatment  because  of  sexual  side­
`effects.  Importantly,  50%  of  the  sample  ‘never  or 
`infrequently’ spoke about sexual functioning with 
`their  primary  mental  healthcare  providers,  and 
`80% of the women with sexual side­effects had not 
`discussed  sexual  dysfunction  with  their  mental 
`healthcare providers. 
`
`Illness beliefs and knowledge of medication
`
`Concepts of health and disease are important factors 
`in  adherence  in  mental  health  (Kelly  et al,  1987). 
`Patients’ understanding of their condition and its 
`need for treatment is positively related to adherence, 
`and  in  turn  adherence,  satisfaction  and  under­
`standing are all related to the amount and type of 
`information given. Studies have shown that patients 
`who  understand  the  purpose  of  the  prescription 
`are twice as likely to collect it than those who do 
`not understand (Daltroy et al, 1991). Awareness of 
`treatment accounted for 20% and 34% of variation 
`in subjective responses (satisfaction with medication 
`and tolerability) to atypical and typical antipsychotic 
`drugs in one recent report (Ritsner et al, 2004). 
`It is often assumed that patients understand a rea­
`sonable amount about their illness, but how often 
`is this assumption tested? In a classic report, Joyce 
`et al (1969) demonstrated that patients were unable 
`to recall half of the information given to them by 
`their physician. Two­thirds of individuals recently 
`discharged from hospital did not know what time to 
`take medication and less than 15% recalled the com­
`mon side­effects (Makaryus & Friedman, 2005). 
`Preliminary  work  suggests  that  figures  in  psy­
`chiatry are comparable. Only 1 in 10 of those tak­
`ing clozapine are aware of potential haematological 
`risks (Angermeyer et al, 2001). Of those prescribed 
`antipsychotics most do not feel involved in treatment 
`decisions and state that they take medication only 
`because they are told to (Gray et al, 2005). Patients 
`typically leave the clinic with a poor understanding 
`of the rationale for therapy (Weiden et al, 2004a). In 
`one study, two­thirds of psychiatric in­patients did 
`not understand why they were taking medication, 
`and the vast majority could not be said to have given 
`informed consent to their treatment (Brown et al, 
`2001). This finding was consistent in both detained 
`and informal in­patients. Similarly, many patients 
`misunderstand prescription instructions. Col et al 
`(1990) found that 50% of patients with depression be­
`lieved they did not need their antidepressants when 
`they began to feel better or that the medication could 
`be taken as required. The UK public campaign Defeat 
`Depression revealed that many people were wary 
`of taking antidepressants because they believed that 
`individuals with depression should ‘pull themselves 
`
`Why don’t patients take their medicine?
`
`together’ and more than three­quarters believed that 
`the medications are addictive (Paykel et al, 1998). In a 
`small sample with mood disorder or schizophrenia, 
`Adams & Scott (2000) found that two components of 
`the health beliefs model (perceived severity and per­
`ceived benefits) and two modifying factors (dysfunc­
`tional attitudes and locus of control beliefs regarding 
`health) differentiated significantly between highly 
`adherent and poorly adherent individuals. Although 
`little work has examined how illness beliefs (good 
`and bad) are formed, previous bad experiences are 
`an important predictor of future non­attendance and 
`non­adherence (Gonzalez et al, 2005). 
`
`Clinician factors
`The doctor–patient relationship
`
`The importance of good communication between 
`patient  and  health  professional  is  increasingly 
`acknowledged in relation to adherence (Stevenson
`et al, 2004). At its essence this means forging a joint 
`therapeutic agreement with full patient involvement. 
`This  is  a  two­way  process  in  which  willingness 
`to  discuss  mental  health  issues  with  a  doctor  is 
`predicted largely by the perceived helpfulness of 
`and trust in that doctor (Wrigley et al, 2005). This can 
`be quantified using a tool that measures therapeutic 
`alliance perceived by patient or clinician. 
`In practice, the process of making a joint thera­
`peutic  plan  is  often  abbreviated.  Doctors  tend  to 
`overestimate the amount of information they have 
`given to patients (Makoul et al, 1995). At the same 
`time, patients often misunderstand medical words. 
`According  to  one  study  (Thompson  &  Pledger, 
`1993), 22% of patients have difficulty with the word 
`‘symptom’, 38% with the word ‘orally’ and 76% with 
`the word ‘stroke’.
`Patients who are very unwell or without insight 
`are unlikely to tolerate an extensive dialogue about 
`possibilities. Yet collaborative decision­making has 
`been shown to be a consistent predictor of health 
`outcomes. Kaplan et al (1996) examined participatory 
`decision­making style on a three­item scale. Higher 
`scores  for  clinicians’  participatory  styles  were 
`associated with greater patient satisfaction and less 
`likelihood of changing doctor. Bultman & Svarstad 
`(2000) conducted an impressive study on reasons 
`for missed antidepressants. They found that 25% of 
`clients were not satisfied with their medication and 
`82% reported missing doses or stopping treatment 
`earlier than recommended. Path analysis showed 
`that patients with more positive beliefs about the 
`treatment are more likely to attend for follow­up and 
`are more satisfied with treatment after attempting 
`medication  use.  Physicians’  communication  style 
`during  follow­up  and  client  satisfaction  were 
`
`Advances in Psychiatric Treatment (2007), vol. 13. http://apt.rcpsych.org/
`
`339
`
`

`
`Mitchell & Selmes
`
`both predictive of better medication adherence. A 
`collaborative communication style by the clinician 
`enhanced  client  knowledge  of  the  medication, 
`improved satisfaction with medication and improved 
`reliability of medication use. 
`
`Predictors of missed medication
`in specific psychiatric disorders
`Depression and antidepressants
`
`The  adherence  habits  of  those  prescribed  anti­
`depressants  have  recently  been  summarised 
`elsewhere (Mitchell, 2006b). Roughly 10% of patients 
`prescribed antidepressants fail to pick up their first 
`prescription and about a third collect only the initial 
`(typically 4­week) prescription. Of those who start 
`medication,  non­adherence  rates  increase  with 
`time (Bultman & Svarstad, 2002). In a study of 200 
`patients attending 14 family doctors in five different 
`practices, non­adherence rates were 16% at week 
`one, 41% at week two, 59% at week three and 68% 
`at week four (Johnson, 1981). In those on long­term 
`maintenance  treatment,  discontinuation  rates  for 
`selective serotonin reuptake inhibitors (SSRIs) are 
`above 70% (Mullins et al, 2005). Of all those who 
`discontinue medication, 60% have not informed their 
`doctor by 3 months and a quarter have not done 
`so by 6 months (Maddox et al, 1994; Demyttenaere
`et al, 2001). This is sometimes referred to as covert 
`non­adherence.
`
`Predictors of missed antidepressant medication
`
`Sirey et al (2001) found that perceptions of stigma 
`about depression at the start of treatment predicted 
`subsequent medication adherence. Aikens et al (2005) 
`asked 81 primary care patients given maintenance 
`antidepressant medications about their adherence. 
`Variation in adherence could be primarily explained 
`by the balance between patients’ perceptions of need 
`v. perceptions of harm, with adherence being lowest 
`when  the  perceived  harm  of  the  antidepressant 
`exceeded  the  perceived  need.  Two  consistent 
`predictors of stopping medication are feeling better 
`and adverse effects. About 35% of patients stop after 
`feeling better at 3 months (Maddox et al, 1994) and 
`55% stop after feeling better at 6 months (Demyttenaere
`et al,  2001).  Ayalon et al  (2005)  examined  both 
`intentional and non­intentional discontinuation of 
`antidepressants. After controlling for ethnicity and 
`medication  type,  intentional  non­adherence  was 
`associated with concerns about the side­effects of 
`the medication and the stigma associated with taking 
`antidepressants. Unintentional non­adherence was 
`associated with greater cognitive impairment. In a 
`study of adherence to SSRIs, Bull et al (2002a) found 
`
`predictors  of  discontinuation  to  be  length  of 
`treatment, low­dose prescription and the occurrence 
`of one or more moderately or extremely bothersome 
`adverse effect. If the adverse effect had been discussed 
`in advance then the likelihood of discontinuation 
`was halved. This is consistent with the finding that 
`individuals with depression generally want more 
`information  about  their  condition  than  they  are 
`offered and want to be involved in decision­making 
`(Garfield et al, 2004). In depression, as in other areas, 
`the  more  information  that  is  given  the  better  is 
`adherence (Maidment et al, 2002).
`
`Mania and mood stabilisers
`Studies  appear  to  support  the  observation  that 
`one­quarter to one­third of patients maintained on 
`lithium are poorly adherent (Cochran, 1986). About 
`a third of those prescribed lithium report that their 
`adherence behaviour is poor and in a similar number 
`sub­optimal serum levels indicate poor adherence 
`(Scott & Pope, 2002). Johnson & McFarland (1996) 
`performed  a  6­year  longitudinal  cohort  study  to 
`determine  patterns  of  lithium  use  in  a  large  US 
`‘health  maintenance  organisation’.  Lithium  users 
`took the drug on an average of 34% of the days. As 
`in other areas, more patients are partially adherent 
`than entirely non­adherent (Colom et al, 2000). 
`In contrast to the above results, MacLeod & Sharp 
`(2001) found much higher rates of adherence, but all 
`of their sample were attending a lithium maintenance 
`clinic, which may have influenced this finding.
`
`Predictors of missed mood stabiliser medication
`
`In  the  handful  of  studies  that  have  examined 
`adherence predictors in bipolar disorders, adverse 
`effects  and  insight  are  common  themes.  In  one 
`study 61% of patients prescribed lithium reported 
`slight or moderate side­effects compared with 21% 
`on  carbamazepine  (Greil  &  Kleindienst,  1999). 
`Recently Bowden et al (2005) reported premature 
`discontinuation rates of about 70% during a 1­year 
`trial  of  lithium,  placebo  and  divalproate:  23.2% 
`of those initially dysphoric at entry who received 
`lithium  maintenance  therapy  and  17.1%  of  those 
`treated  with  divalproate  maintenance  therapy 
`discontinued  prematurely  because  of  side­effects 
`compared with 4.8% of those treated with placebo. 
`Regarding illness severity and insight, Yen et al (2005) 
`found that insight at baseline predicted concordance 
`up to 1 year later.
`
`Schizophrenia and antipsychotics
`Perhaps the most striking data come from studies 
`of full discontinuation by people with schizophrenia. 
`
`340
`
`Advances in Psychiatric Treatment (2007), vol. 13. http://apt.rcpsych.org/
`
`

`
`Even in highly monitored randomised controlled 
`trials, discontinuation rates can be greater than 50%. 
`At  least  five  large­scale  studies  have  shown  that 
`adherence with both old and new antipsychotics is 
`poor in the long term. For example, Rosenheck et al 
`(1997)  found  that  68%  of  patients  treated  with 
`haloperidol and 43% of those treated with clozapine 
`had discontinued medication before the end of a 
`1­year trial. In the 18­month Clinical Antipsychotic 
`Trials for Intervention Effectiveness (CATIE) study 
`(Lieberman et al, 2005) a remarkable 74% of patients 
`discontinued medication prematurely. The most com­
`mon reasons for discontinuation were patient choice, 
`lack of effect or intolerability of side­effects. 
`Outside  of  randomised  controlled  trials  many 
`groups have looked at more subtle forms of poor 
`concordance, including under­ and excessive dosing 
`by people with schizophrenia. Seven studies have 
`recruited more than 1000 patients (these are reviewed 
`by Velligan et al, 2006). Rijcken et al (2004) found that 
`33% of the patients in their sample had sub­optimal 
`prescription  renewals  (less  than  90%),  56%  had 
`good concordance and 11% had excess prescription 
`renewals. Weiden et al (2004b) found that 90% of a 
`sample of 675 individuals with schizophrenia had 
`some degree of partial adherence and on 36% of the 
`study  days  someone  had  run  out  of  medication. 
`Patient surveys suggest that almost half (44%) have at 
`some point stopped taking their medication without 
`agreement of their doctor (Hogman & Sandamas, 
`2000).
`
`Predictors of missed antipsychotic medication
`
`There  is  a  concern  that  people  diagnosed  with 
`schizophrenia  are  infrequently  involved  in  treat­
`ment decisions and sometimes not even told their 
`diagnosis  (Bayle et al,  1999).  One  patient  survey 
`(Gray et al, 2005) reports that most of the participants 
`prescribed  antipsychotics  did  not  feel  involved 
`in  treatment  decisions  and  had  not  been  given 
`written information about their treatment, warned 
`about side­effects or offered non­pharmacological 
`alternatives.
`Several  cross­sectional  studies  link  severity  of 
`psychopathology to medication non­adherence (Van 
`Putten et al, 1976; Pan & Tantam, 1989). Grandiose 
`thoughts and persecutory thoughts may carry partic­
`ular risk (Van Putten et al, 1976). Low insight predicts 
`non­adherence and, importantly, improvements in 
`insight are often accompanied by improvements in 
`concordance (Rittmannsberger et al, 2004). However, 
`low insight may be linked to poorer cognitive func­
`tion and one study suggests that during the first 
`year of treatment, patients with poorer premorbid 
`cognitive functioning are more likely to discontinue 
`(Robinson et al, 2002). 
`
`Why don’t patients take their medicine?
`
`Lack of insight has been extensively examined 
`but other factors have been relatively overlooked. 
`Perkins  (2002)  reviewed  articles  published  up 
`to  December  2002.  Correlates  of  poor  adherence 
`included patients’ beliefs about their illness and the 
`benefits of treatment (insight into illness, belief that 
`medication can ameliorate symptoms), perceived 
`costs  of  treatment  (medication  side­effects),  and 
`barriers to treatment (ease of access to treatment, 
`degree  of  family  or  social  support).  More  recent 
`studies  suggest  that  medication  concerns  are  of 
`prime importance. In a sample of 213 patients with 
`schizophrenia, Lambert et al (2004) found that those 
`presenting  with  side­effects  and  those  with  past 
`experience of side­effects had a significantly more 
`negative  general  attitude  toward  antipsychotics. 
`Loffler  et al  (2003)  conducted  a  study  in  which 
`307 people with schizophrenia were asked about 
`their reasons for antipsychotic adherence or non­
`adherence, using the Rating of Medication Influences 
`scale. A positive relationship with the therapist and 
`a  positive  attitude  of  significant  others  towards 
`antipsychotic treatment contributed to adherence. 
`Reasons for non­adherence were lack of acceptance 
`of the necessity for pharmacological treatment and 
`lack of insight.
`
`Drugs and alcohol
`
`Batel et al (2004) examined risk factors of early drop­
`out during induction of high­dose buprenorphine 
`substitution therapy in 1085 individuals addicted to 
`opiates. Younger age, lack of social support and partial 
`access  to care (lack of health insurance, previous 
`contact  with  the  prescriber)  were  significantly 
`associated with early drop­out. Herbeck et al (2005) 
`looked at predictors in those with dual diagnosis. 
`Patients with treatment adherence problems were 
`significantly more likely to have personality disorders, 
`lower Global Assessment of Functioning scores, and 
`medication side­effects than those who adhered to 
`treatment.  In  a  study  of  after­care  attendance  by 
`severely substance­dependent residential treatment 
`clients, Sannibale et al (2003) found that younger, 
`male, heroin­dependent clients with poly­drug use 
`who  had  refused  opioid  pharmacotherapy  were 
`more likely to drop out of treatment and to relapse 
`early following treatment.
`
`Discussion
`
`Rates of non­adherence with psychotropic medication 
`are  difficult  to  summarise  because  they  vary  by 
`setting, diagnosis and type of adherence difficulty. 
`The overall weighted mean rate of non­adherence, 
`calculated in a sample of 23 796 patients with psychosis 
`
`Advances in Psychiatric Treatment (2007), vol. 13. http://apt.rcpsych.org/
`
`341
`
`

`
`a quarter of those discontinuing antidepressants or 
`antipsychotics  cite  adverse  effects  as  the  reason 
`(Maddox et al, 1994). In other words, only a minority 
`of those experiencing significant adverse effects will 
`discontinue medication without advice. 
`A  substantial  minority  (about  a  quarter  to  a 
`half) do not tell their doctor after stopping or inter­
`rupting medication (Hogman & Sandamas, 2000; 
`Demyttenaere et al, 2001). In part this may be fear 
`of rejection or being disbelieved, or embarrassment 
`about  discussing  adverse  effects  such  as  weight 
`gain or sexual problems (Zimmermann et al, 2003; 
`Weiden et al, 2004a). A contributing factor may be 
`that the frequency of side­effects such as these is 
`greatly underestimated by doctors (Smith et al, 2002; 
`Roose, 2003). About half of people who stop their 
`medication after taking it for more than a month 
`do so intentionally (Barber et al, 2004). Even when 
`patients have discontinued medication unexpectedly, 
`studies in general medicine suggest that more than 
`half will offer some kind of rational explanation. 
`
`Box 3 Simple strategies to improve
`concordance
`
`Basic communication
`Establish  a  therapeutic  relationship  and 
`trust
`Identify the patient’s concerns
`Take into account the patient’s preferences
`Explain  the  benefits  and  hazards  of  treat­
`ment options
`Strategy-specific interventions 
`Adjusting  medication  timing  and  dosage 
`for least intrusion
`Minimise adverse effects
`Maximise effectiveness
`Provide 
`support,  encouragement  and 
`follow­up
`Reminders
`Consider adherence aids such as medication 
`boxes and alarms
`Consider  reminders  via  mail,  email  or 
`telephone
`Home visits, family support, counselling
`Evaluating adherence
`Ask about problems with medication
`Ask specifically about missed doses
`Ask about thoughts of discontinuation
`With  the  patient’s  consent,  consider  direct 
`methods: pill counting, measuring serum or 
`urine drug levels
`Liaise  with  general  practitioners  and 
`pharmacists regarding prescriptions
`
`• • • • • • •
`
`•
`
`• • • •
`
`• • • •
`
`Mitchell & Selmes
`
`from a systematic literature review was 25.78% (Nose 
`et al,  2003).  Complete  discontinuation of  medica­
`tion is thought to lead to about one in ten hospital 
`admissions and one in five nursing home admissions 
`(Sullivan et al, 1990). Fenton et al (1997) found that 
`non­adherent individuals with schizophrenia have 
`a  3.7­fold  greater  risk  of  relapse  than  those  who 
`are adherent over 6–24 months. Where medication 
`(or  appointments)  are  missed  for  predominantly 
`illness­related reasons such as lack of insight, there 
`is a particularly high risk of readmission. Yet illness 
`severity probably accounts for a minority of cases 
`of poor adherence in the community (Maddox et
`al, 1994). Further, the impact may be ameliorated 
`if patients who have further symptoms seek help. 
`Unfortunately, adverse experiences with medication 
`may prejudice willingness to attend in the future 
`(Gonzalez et al, 2005). 
`
`Potential solutions
`A  comprehensive  summary  of  interventions  that 
`may reduce poor adherence is beyond the scope of 
`this article. Advanced strategies such as adherence 
`therapy  and  depot  medication  are  reviewed  by 
`O’Ceallaigh & Fahy (2001) and Nadeem et al (2006). 
`Many  of  the  simpler  interventions  available  to 
`enhance adherence are listed in Box 3.
`Barriers to healthcare are a reversible cause of poor 
`medication and appointment adherence. Common 
`factors  that  reliably  influence  adherence  include 
`patient expectation and knowledge (perception of 
`benefits and hazards of therapy), involvement in 
`medical decisions, availability of social support, and 
`complexity and duration of the prescribed regime. 
`In the past it has been common to blame the indi­
`vidual for discontinuing medication or dropping out 
`of treatment against medical advice (Demyttenaere, 
`1998).  In  addition,  patients  who  do  not  follow 
`medical  advice  are  likely  to  be  discharged  early, 
`compounding the problem. In the future this may 
`be seen as inadequate care. Not too long ago it was 
`considered unusual to give the patient information 
`about their condition or, in some cases, even to reveal 
`the  truth