Article Text

Download PDFPDF

Medication discrepancies revealed by medication reconciliation and their potential short-term and long-term effects: a Norwegian multicentre study carried out on internal medicine wards
  1. Niklas Nilsson1,2,
  2. Marianne Lea1,2,
  3. Yvonne Lao1,2,
  4. Katherine Wendelbo2,3,
  5. Gløer Gløersen2,4,
  6. Morten Mowé5,6,
  7. Hege Salvesen Blix2,7,
  8. Kirsten K Viktil2,8
  1. 1Department of Pharmaceutical Services, Oslo Hospital Pharmacy, Hospital Pharmacies Enterprise, Oslo, Norway
  2. 2School of Pharmacy, University of Oslo, Blindern, Oslo, Norway
  3. 3Levanger Hospital Pharmacy, Central Norway Hospital Pharmacy Trust, Levanger, Norway
  4. 4Hospital Pharmacy Tønsberg, Hospital Pharmacies Enterprise, Tønsberg, Norway
  5. 5Department of General Internal Medicine, Oslo University Hospital, Oslo, Norway
  6. 6Institute of Clinical Medicine, University of Oslo, Blindern, Oslo, Norway
  7. 7Department of Pharmacoepidemiology, Norwegian Institute of Public Health, Oslo, Norway
  8. 8Diakonhjemmet Hospital Pharmacy, Oslo, Norway
  1. Correspondence to Niklas Nilsson, Department of Pharmaceutical Services, Oslo Hospital Pharmacy, Kirkeveien 166, Oslo 0450, Norway; niklas.nilsson{at}


Purpose To investigate the extent of medication discrepancies (MDs) revealed by medication reconciliation (MR) and to assess the potential clinical relevance of the MDs for the patients in a short-term and long-term perspectives.

Methods Patients ≥18 years admitted to five internal medicine wards were included in this prospective study. MDs between the medication list obtained by physicians at hospital admission and medication list obtained by a structured MR process by pharmacists were identified and assessed for clinical relevance by an expert team. Clinical relevance was assessed in two ways as (a) if they were not acted upon during the hospital stay (short term) or (b) if they persisted after discharge from the hospital (long term).

Results In total 262 patients, age 19–98 (SD 18.94, mean 73.4 years), 58.8% female, were included. 79.4% of the patients had at least one MD with a mean of 3.2 MDs/patient. 80.7% of the MDs were discussed with the physician, and 95.5% of these were acted upon. Of the 438 MDs evaluated by the expert panel, 35.2% and 71.2% were assessed to be of moderate, major or extreme clinical relevance in the short-term and long-term perspectives, respectively.

Conclusions By using a structured approach, MDs were identified for 80% of the patients and the majority of the MDs were evaluated to possibly harm the patient in a long-term perspective. The results emphasise that structured MRs may improve patient safety.

Trial registration number 2011/542.


Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


Inaccurate medication lists is a huge problem, especially when changing levels in the healthcare system.1–10 The Institute for Healthcare Improvement claims that poor communication or lack of communication of patients’ medication treatment at transition points is the cause of 50% of all medication errors and 20% of all adverse drug reactions in hospitals.11 An inaccurate or incomplete medication list obtained at hospital admission may lead to interruptions of regular drugs and/or initiation of inappropriate drug treatments during the hospital stay and this could persist in the patient's medication list after discharge. This might lead to increased morbidity and mortality and also increased costs due to longer hospital stays or readmissions. WHO has recognised medication reconciliation (MR) as a method to obtain a complete medication list.12 Tam et al1 presented in a review article that up to 67% of patients have at least one medication discrepancy (MD) in the medication list when admitted to hospital and 11%–59% of the discrepancies were clinically relevant. A Danish study found that 69% of the patients had one or more MDs at admission to hospital and 18% of these discrepancies were considered to be potentially serious.8 MR should be addressed by focusing on the outcomes of the discrepancies, that is, the clinical relevance of the discrepancies, both in the short-term and long-term perspectives. To our knowledge, only one earlier study has explored the clinical relevance of MDs in the long-term perspective.13

The objective of this study was to investigate the extent to which the individual patient’s medication list recorded on admission to Norwegian hospitals differs from the patient’s actual drug regimen prior to admission. Furthermore, the objective was to assess the potential clinical relevance of the identified MDs both short term (ie, during the few days of a hospital stay) and long term (ie, maintained after discharge).

Materials and methods

Patients and design

This prospective multicentre study was approved by the Norwegian Regional Committee for Medical and Health Research Ethics and the Privacy Ombudsman at each hospital. It was performed on two geriatric wards, one infection ward, one nephrology ward and one general internal medicine ward, in five hospitals in South-Eastern Norway during the period of August to November 2011. Patients ≥18 years with signed informed consent forms were consecutively included in the study. Terminal ill patients and readmitted patients were excluded. Patients isolated due to infection, and those who did not understand Norwegian or English were also excluded.

Data collection

MR was conducted by a clinical pharmacist within 48 h after the patient's admission to the ward, using the integrated medicines management (IMM) model.14 All pharmacists involved in the study underwent the same training for executing an MR and used the same forms to record data. The IMM model has been translated and adapted to the Norwegian setting and was pilot tested at each ward. The following were recorded from the patient's electronic medical record: age, gender, relevant medical history and the reason for the hospitalisation. The level of care was also recorded (if the patient received help with handling of their drugs before admission or if the patient received their drugs in a multidose system). In a multidose system the individual patient's medications are packed by a dispensing pharmacy in labelled plastic bags; each bag contains all the tablets that should be taken at each dosing time. The identified drugs were recorded according to the Anatomical Therapeutic Chemical system (ATC system)15 with its drug name, generic name, formulation, strength, time of dose and dosage.

The standard method for obtaining a patient's medication list at on admission to the hospital is made by the physician who develops a handwritten medication chart either by asking the patient and/or by using the level of care who referred the patient to the hospital. MR were performed by the clinical pharmacist by a structured interview with the patient using a standard interview form, including a checklist to discover which medications, including the actual dosages were used by the patient prior to admission. The checklist is essential to reveal drugs that patients often forget to mention and also the physician forget to ask about at admission. Typical questions included in the checklist were regarding the use of eye-drops, contraceptives and other hormones, inhalations as asthma medicines, non-prescription medication, weekly/monthly dosed drugs and specific drug types, for example, painkillers, cardiovascular drugs and hormones, and also natural drugs and dietary supplements. If the pharmacist did not get satisfactory information from the patient, the pharmacist collected information from additional sources, for example, the patient’s general practitioner, the community pharmacies or the patients’ relatives. The clinical pharmacist obtained for patients not handling their own drugs prior to admission a pre-admission medication list from the person or healthcare level that was responsible for handling the patients’ drugs prior to the admission; nursing homes, district nurses or the patients relatives.

Medication discrepancies

The medication list obtained by the clinical pharmacist by the structured interview of the patient and/or collecting data from relevant level of care was compared with the medication list documented by the physician on admission to the hospital. MDs between these two lists were recorded. Intentional changes in prescribing, as described in the patient journal or the handwritten patient chart, were not considered as MDs. Otherwise the pharmacists did not question each physician in retrospect if an MD was intentional or not. MDs were registered for compounds classified as drugs by ATC system. Herbal drugs and dietary supplements were also recorded but not assessed with regard to discrepancies between lists; however, information about the use of these compounds if considered relevant were discussed with physicians. MDs were categorised into six categories: 1, commission of drug (prior to admission the patient did not use the drug prescribed in the admission medication list written by the physician); 2, omission of drug (prior to admission the patient used a drug, ie, not prescribed in the admission medication list); 3, incorrect strength of drug (discrepancies between the lists, eg, 5 mg is written in the medication chart but the patient uses 10 mg); 4, incorrect drug formulation; 5, incorrect dose (eg, 1.25 mg×1 is written in the medication chart but the patient uses 1.25 mg×2) and 6, incorrect administration time of dose.

All MDs revealed by the pharmacist were recorded and the pharmacist made individual assessments if he/she should intervene, that is, discuss the MDs with the attending physician on the ward. MDs brought up for discussion with the physician and agreed upon for action were recorded.

Clinical relevance

An expert panel consisting of three experienced clinical pharmacists and one senior geriatrician retrospectively evaluated the potential clinical relevance of the identified MDs. Every second patient was selected for this evaluation, but if this patient did not have any MD, the next patient with an MD was selected. The expert panel used the following information for each patient when assessing the clinical relevance: medication list before and after reconciliation, age, gender, reason for hospitalisation, former and current diseases and the level of care before admission. Members of the panel individually assessed the clinical relevance of the MDs and thereafter held a consensus meeting. Clinical relevance was assessed in two situations for each MD: if the MDs persisted for short term and if it persisted for long term. The short-term effect was defined as of clinical relevance if the discrepancy would not be discovered during the few days of the hospitalisation. The long-term effect was defined as of clinical relevance if the MDs would continue after hospital discharge and be part of the patient's post-discharge medication list. Even though this is presumed discrepancies, this post-discharge medication list usually will be in use at least until the patient sees their GP, which could be several weeks after discharge from the hospital. Further, it is not definitely that the GP reveals the MDs, but assume they are intended medication changes made at the hospital. Regardless persisting weeks, months or years, this would be long-term MDs.

The clinical relevance of the MDs was graded as follows: (1) extreme clinical importance included MDs needing interventions to prevent death or severe or irreversible detrimental effects; (2) major clinical importance included discrepancies needing interventions to prevent major or reversible detrimental effects, or lack of therapy in conditions where evidence-based options were available; (3) moderate clinical importance included discrepancies where interventions may result in moderate benefit for the patient and (4) minor clinical importance included discrepancies of little clinical relevance for the patient, such as minor time adjustment of dosage regimens.16


A database was established and the data analysed with SPSS V.20.0 for Windows. Descriptive statistics are given as means and SDs for continuous variables and as frequencies for categorical variables. To test for differences between groups of patients with and without discrepancies, an independent samples t test was used for continuous variables and Pearson's χ2 tests was used for categorical variables. p Values of <0.05 (p<0.05) were considered statistically significant.


Patient characteristics and MDs

A total of 262 patients were included in the study, finding a total of 1402 drugs before reconciliation and 1582 drugs after reconciliation. Patient characteristics are given in table 1.

Table 1

Demographic characteristics, number of drugs and number of discrepancies in 262 patients admitted to internal medicine wards

A total of 662 MDs were revealed, and 208 of the patients (79.4%) had at least one MD, a mean of 3.2 MDs per patient among those having MDs. Patients with MDs used significantly more drugs both before and after reconciliation, than patients without MDs (table 1). Figure 1 shows the relationship between the number of drugs and the number of MDs, r2=0.71. Type and number of MDs are shown in table 2. Most of the discrepancies, 90.6% (n=600), were associated to regularly used drugs. There were no significant difference between patients with one or more discrepancies and patients with no discrepancies regarding age, gender, if they had help with medication or not, if admitted from home or not and if they had drugs delivered in a multidose system or not.

Table 2

Distribution of the 662 discrepancies revealed by medication reconciliation in 262 patients regarding type of discrepancy and drugs most involved

Figure 1

Frequency of discrepancies per patient in relation to number of drugs used on admission (ie, after medication reconciliation). r2=0.7103, n=262 patients.

The pharmacist discussed 80.7% (534) of the MDs with the attending physicians. The physicians agreed to act upon 510 (95.5%) of these MDs and immediate adjustments were made in the medication lists for 423 MDs (82.9%). The rest of the MDs (128 discrepancies; 19.3%) were not brought up for discussion either because the pharmacist assessed these discrepancies to be of less importance or they had been solved already. Further, 86 patients (33%) used 157 different types of herbal drugs, vitamins or natural drugs.

Of the 208 patients with discrepancies, 132 patients (63.5%) were selected to have their MDs evaluated by the expert panel with regard to clinical relevance. The total number of MDs in these patients was 438 (of the total 662 discrepancies), with an average of 3.3 MDs per patient, an average of 5.9 drugs on admission and an average age of 75.0 years. The patients who were not evaluated by the expert panel (76 patients) had an average of 2.9 discrepancies, 5.6 drugs on admission and an average age of 75.5 years.

Table 3 summarises the expert panel's assessments of clinical relevance of the MDs during the hospital stay (short-term effect) and if the discrepancies were maintained after discharge from the hospital (long-term effect). On short-term perspective, 35.1% of the 438 MDs evaluated were assessed to be of moderate, major or of extreme importance regarding clinical relevance, whereas considering long-term perspective a total of 71.2% of the MDs were evaluated to be in these categories. An example of a discrepancy that was evaluated to be of extreme importance regarding clinical relevance during the hospital stay was when a digitalis glycoside was dosed 10 times higher than intended.

Table 3

Clinical relevance assessed in the 438 selected discrepancies in 132 patients during 1 week of hospital stay (short term) or after discharge (long term) and the most frequent drugs involved

Of those discrepancies presuming to persist after discharge (long-term perspective), 21 MDs (4.8%) in 15 patients (11.4%) were considered to be of extreme importance regarding clinical relevance, that is, every ninth patient (15 of 132 patients) could potentially have an irreversible effect if their discrepancies were to persist after discharge. Examples were omission of eye-drops treating glaucoma and omission of antiviral treatment for HIV. The most frequent drug group with discrepancies was glaucoma eye-drops. Of the 50 patients who used glaucoma eye-drops, 33 had at least one discrepancy in the medication list regarding the use of glaucoma eye-drops.


This multicentre study from Norwegian hospitals revealed that nearly 80% of the patients had at least one MD between the medication list obtained by the physician using the standard method on admission and the medication list obtained by the clinical pharmacist performing a structured interview including collecting data from relevant sources. This is in accordance with other studies1–7 and shows through the literature that obtaining accurate medication lists on admission to a hospital is a huge challenge. Further, our findings showed that in the short-term perspective about 6% of the MDs were assessed to be of potentially extreme importance or of major clinical relevance, whereas in the long-term perspective this figure was increased to nearly 35%. This highlights the importance of performing structured MR when patients are admitted to hospital.

The use of different scales and methods to evaluate the clinical relevance of the discrepancies revealed by MR, make the comparison of our results with other studies difficult. However, a study from Canada where a three-step scale was used to categorise clinical relevance in 151 patients, found that 38.6% of the discrepancies had the potential to cause moderate to severe harm to the patient during the hospital stay.4 This is in line with our results where 35.2% of the discrepancies were evaluated to be of extreme, major or moderate importance to clinical relevance in the short-term perspective. Two French studies showed that 20.8% and 42% of the MDs, respectively, had the potential to require increased monitoring or intervention to preclude harm, and 6.4% and 12%, respectively, had the potential to harm the patient.17 ,18 Neither the Canadian nor the French studies did assess the potential of the MDs to harm the patient in long-term perspective.

To our knowledge only one study has reported on long-term consequences of MDs. Gleason et al13 found an increase in severity from 22% in the short-term perspective to 59% in the long-term perspective. Our study revealed slightly higher number 35.1% and 71.2%, respectively. Gleason et al used a severity scale from A to I to predict clinical relevance of discrepancies. In the long-term perspective Gleason found that 3 of 69 discrepancies (4.3%) could result in permanent harm or in death (categories G–I in severity scale) and our study showed 4.8%. Gleason analysed 69 discrepancies and our study assessed 438 discrepancies. Our study shows the potential consequences of the discrepancies persist over time and that they become more serious if not detected and taken care of.

To illustrate the different assessment of MDs, in the short- and long-term, omission of glaucoma eye-drops during hospital stay was evaluated to be of moderate clinical relevance, whereas in the long-term perspective the expert group evaluated this discrepancy to be of extreme important clinical relevance. The appraised number of extreme important clinical relevant discrepancies increased from 1 (0.2%), in the short term, to 21 (4.8%) in the long term. Furthermore, every ninth patient could have irreversible detrimental effects in a long-term perspective if their discrepancies were not revealed. This means that the discrepancies in general were appraised to cause less harm in the short-term than in the long-term perspective. Further, the same medication could have MDs of different clinical relevance in different patient cases due to patient characteristics, that is, illness, age and/or current disease. An example was a patient who used paracetamol 1 g×4 against pain prior to admission; however, in the admission note it was only recorded paracetamol 1 g as needed. This MD was evaluated of higher clinical relevance than when a patient rarely used paracetamol at home and it was not recorded paracetamol at all in the admission note.

Our study showed that omission of a drug from the medication list (53% of the discrepancies) was the most common discrepancy, and this is in accordance with other studies.4 ,7 ,17 ,19 We found, for example, that weak analgesics and drugs for constipation were the most omitted drugs. Unroe et al3 found that cardiovascular drugs were the most frequent drug group associated with MDs. These omissions could be caused by incomplete communication between caregivers.

Nearly two out of three patients using eye-drops for glaucoma had at least one discrepancy regarding the use of glaucoma eye-drops. This emphasises the need for a systematic approach to conduct MR and the importance to include direct questions about drugs that patients do not think of as typical medications, or drugs they think do not matter to mention when they are admitted to an internal medicine ward. Our structured form including the checklist was essential to reveal the use of these types of drugs by the patients.

The finding of a correlation, between the number of drugs on admission before reconciliation and MDs, is in line with Gleason et al9 findings; however, in contrast to Unroe et al3 who did not find a correlation between the number of drugs and MDs. In our study, MDs were also revealed in patients registered not to use drugs on admission by the physician, that is, omission discrepancies regarding antiviral treatment for HIV. This suggests that using the number of drugs to prioritise which patients to reconcile can lead to potential detrimental effects and hence reduced patient safety. Our study did not find any correlation between age, gender, level of care and the extent or type of discrepancies, which is in contrast to Hellström et al5 who found that patients without any care more often had discrepancies in their medication lists. Midlöv et al7 found more discrepancies at discharge from the hospital in patients with drugs delivered in a multidose system.

The fact that discrepancies of extreme important clinical relevance in long term were revealed in every ninth patient of those evaluated is worrying. Therefore, it should be mandatory to perform a structured MR as an important contribution to patient safety. This could save harming patients and/or saving lives resulting in savings for the healthcare system. As discussed it is important to have a structured approach when performing MR. In our study, we found that almost all revealed discrepancies were accepted as an MD and acted upon by the physicians. This supports the suitability of the structured method to perform MR and also that clinical pharmacists should be a part of the treatment team on hospital wards to increase patient safety.

The strength of this study is the prospective multicentre design. However, it was only performed on internal medicine wards, and as such the results cannot be generalised to other type of wards. Another strength of this study was that an external team assessed the clinical relevance of the MDs revealed, and that this was assessed both on a short-term and on a long-term perspective. The expert team consisted of both experienced clinical pharmacists and a geriatrician. It was important to have a geriatrician in the team as most of the patients admitted to internal medicine wards in these hospitals are elderly; the mean age in our study was more than 73 years.

This study has some limitations. A limitation is that the assessment of the clinical relevance of the MDs was assessed retrospectively, and every second patient was assessed. However, the patients that did not have their MDs assessed did not differ from the patients assessed with regard to age, number of drugs and number of discrepancies. Another limitation is our definition of long-term MDs; presuming that MDs would persist according to the post discharge medication list. Additionally, the number of MDs could have been overestimated because the pharmacist did not ask the physician whether the MDs were intentional or not.

It is well known that when the patient is transferred to or within the healthcare system, the loss of information due to poor communication could be harmful for the patient,11 and our study has highlighted that the consequences of information loss leads to more potential serious harm to the patient if existing over time. It should be mandatory that each caretaker involved in the patient treatment and also in the handling of medicines should take responsibility for having an updated and correct medication list. An important element is to think of quality assurance, prevention of disease or adverse events. When it comes to prevention in the healthcare system it is well known it is better to prevent than to treat. An example is the standard procedure of giving medicines as secondary prophylaxis after a stroke or heart attack to prevent new events. The same thought should be used when it comes to having an updated and correct medicine list.

To increase patient safety we recommend that priority should be given to implement structured methods for performing MR.


By using a structured approach, MDs were identified for 80% of the patients and the majority of the MDs were evaluated to possibly harm the patient in a long-term perspective. The results emphasise that structured MRs may improve patient safety.

Key messages

  • What is already known on this subject?

  • Medication discrepancies are common and could potentially harm the patient.

  • Medication discrepancies can be due to poor communication between caregivers or because of not having a standardised approach to assure a correct and updated medication list.

  • A structured and standardised method to perform medication reconciliation on admission to hospital could improve patient safety.

  • What this study adds?

  • Medication discrepancies could be of potential harm to the patient during hospital stay (short-term perspective) but could be of potential harm to the patient even more if the discrepancies persist after discharge (long-term perspective).


We would like to thank Hospital Pharmacies Enterprise, South Eastern Norway and Lovisenberg Hospital Pharmacy for all support to conduct this study. You made it possible. Many thanks to clinical pharmacist Anne Katrine Eek from Lovisenberg Hospital Pharmacy, Oslo for her excellent help to evaluate the potential clinical relevance of the identified MDs. Thanks to Wendy Klem for language vetting. We also would like to thank all physicians, nurses and other healthcare staff at Akershus University Hospital, Lovisenberg Diakonal Hospital, Oslo University Hospital and Vestfold Hospital for all their help and dedication.



  • Contributors NN, ML, YL, KW, GG, HSB and KKV designed the protocol. NN, ML, YL, KW and GG collected data. ML, NN and MM evaluated the clinical relevance of the MDs. HSB was responsible for the statistical analysis. MM, HSB and KKV were local supervisors.

  • Competing interests None declared.

  • Patient consent Obtained.

  • Ethics approval This study was approved by the Norwegian Regional Committee for Medical and Health Research Ethics.

  • Provenance and peer review Not commissioned; externally peer reviewed.