Prospective observational study of medication reviews in internal medicine wards: evaluation of drug-related problems

Abstract

Background

The Lund Integrated Medicines Management model offers a systematic approach for individualising and optimising patient drug treatment. Clinical, economical and humanistic outcomes have been shown as well as results from the medication reconciliation process. There is a need also to describe the medication review process.

Objective

To describe the frequency and types of drug-related problems (DRPs) identified during medication reviews and to evaluate the actions of the pharmacists and the physicians regarding the identified DRPs.

Method

Structured medication reviews were conducted by a multi-professional team on top of standard care for 719 patients in two internal medicine wards in a Swedish University Hospital. The medication reviews were studied retrospectively to classify DRPs and actions taken.

Results

A total of 573 (80%) of patients had at least one actual DRP; an average of three DRPs per patient and in total 2164. Wrong drug and adverse drug reaction were the most common types of DRPs. The most frequent medication groups involved in DRPs were drugs for the cardiovascular system and the nervous system and the most frequent substances were warfarin, digoxin, furosemide and paracetamol. The 10 most common medications accounted for 27% of the actual DRPs. Of the identified DRPs, a total of 1740 (80%) were acted on. The three most common types of adjustments made were withdrawal of drug therapy, change of drug therapy and initiation of drug therapy. When the pharmacist suggested an adjustment, the physician implemented 88% (1037/1174) of the recommendations.

Conclusion

DRPs are common among elderly patients who are admitted to hospital. Systematic identification of high-risk medications and common DRP types enables targeting of prioritised patients for medication reviews.

Introduction

The identification, prevention and resolution of drug-related problems (DRPs) have been stated to be the core process of pharmaceutical care, with the aim to improve patient outcome.¹ A DRP, which has been described as “an undesirable patient experience that involves the drug therapy and that actually or potentially interferes with the desired patient outcome”,² may occur for many reasons such as prescribing errors or inadequate monitoring. This may result in adverse drug effects or drug treatment goals not being achieved. In general, elderly patients with multiple diseases and polypharmacy are considered more vulnerable to having a DRP.

DRPs have been shown to cause admission to hospital,^3–7 and a considerable number of these are preventable.^{5 8} Adverse drug reactions, overprescribing and suboptimal drug therapy are common causes of drug-related admissions.⁹ The risk of drug-related hospitalisations is much higher in older adults.¹⁰ DRPs in hospitalised patients are of major concern^{9 11} and may have physical as well as economic implications. Approaches to identify and decrease DRPs are therefore of great interest.

The Lund Integrated Medicines Management (LIMM) model offers a systematic approach for individualising and optimising drug treatment for inpatients¹² and patients in primary care.¹³ A clinical pharmacist is used as a catalyst for improvement, but the whole patient care team is crucial. In previous studies, in-hospital medication reviews have been shown to have a positive effect on DRPs¹⁴ and optimisation on medication appropriateness and rates of drug-related readmission¹² but not regarding readmission in general.¹⁵ However, in order to adjust arisen and prevent future DRPs, there is also a need to describe the process of the medication review and describe the type of DRPs and involved drugs.

The aim of this study was to describe the frequency and types of DRPs identified by pharmacists performing medication reviews and to evaluate the actions of the pharmacists and the physicians regarding the identified DRPs. In addition, we aimed to analyse certain drug groups and substances, and untreated symptoms or diseases associated with actual DRPs.

Methods

Setting and population

This prospective study was conducted in two internal medicine wards at the University Hospital of Lund, Sweden. A LIMM-based clinical pharmacy service, including medication reconciliation, as presented elsewhere,¹⁶ and medication review during the hospital stay, was implemented in January and October 2007, in the respective wards. All patients were eligible for inclusion in the study if they had been admitted to the wards after implementation of the service until the end of the year 2007. The LIMM medication review included two levels: basic and extended. Patients were excluded from the study if they had been discharged or deceased before the pharmacist could conduct a medication review.

There were 22 beds in each ward. The weekday staff in each ward was composed of two junior physicians and two senior physicians, one clinical pharmacist, three nurses, three assistant nurses, one physiotherapist and one occupational therapist. Availability of beds alone was the deciding factor regarding the ward to which a patient was admitted. The wards used the standard hospital electronic health record (EHR) system (Melior; Siemens) that included a computerised physician order entry system.

Medication review process in the LIMM model

The medication review process in the LIMM model is structured and was developed over approximately 5 years and implemented on top of standard care.^{17 18} According to the protocol, the clinical pharmacists should conduct medication reconciliation for all patients. Before the medication review was conducted, patients were allocated to basic or extended medication review. Patients allocated to extended medication reviews were the following: 65 years or older, not considered terminal and not admitted to hospital due to planned examinations, and had at least one risk factor. The risk factors were the following: cognitive failure, heart failure treated with both ACE inhibitors and beta blockers, renal failure with a creatinine clearance of <20 mL/min (Cockroft-Gault equation), or being prescribed digoxin, warfarin, non-steroidal anti-inflammatory drugs, antiarrhythmic drugs, theophylline, fluconazole, tramadol or amiloride. Other patients were allocated to basic medication reviews. The intention was to include all patients. However, patients receiving extended medication reviews were prioritised due to case time or personal constraints.

Weekdays before morning rounds, the clinical pharmacist conducted structured medication reviews using the LIMM form. When conducting the extended medication reviews, the clinical pharmacist monitored all available relevant information on the patients. This was done by using the medication lists in the patient records, health record notes, laboratory test results and other results (eg, blood pressure). The basic medication reviews were conducted using the medication list in the patient records. The clinical pharmacist used the LIMM medication review form to identify DRPs according to specific categories of risk and formulated suggestions for changes to medication treatment. The nine categories of risk were (1) drugs that required therapeutic drug monitoring, (2) less appropriate drug therapy,¹⁹ (3) interchangeable drugs (generic and analogous substitution according to a regional interchangeable list), (4) problems with handling the drugs (eg, swallowing, crushing), (5) interactions involving a drug combination that could require dose adjustment or a drug combination that ought to be avoided, (6) drug or dose not adjusted according to renal function, (7) unclear indication, (8) untreated/not optimal treated indication and (9) suspected adverse drug reaction.

During rounds, the patients’ drug treatment was discussed by multi-professional teams, consisting of physicians, nurses, care providers, the clinical pharmacist and paramedics. The identified DRPs and other clinical opinions served as a basis for any decision to change the drug treatment.

Patients were followed up twice weekly to enable identification of new DRPs and to monitor previously identified problems and the clinical pharmacist attended rounds once to twice weekly for each patient. On average, pharmacists using the LIMM model spend a median 1 hour on each patient performing medication reconciliation at admission and medication review during the hospital stay¹⁷; extended medication reviews take on average 10 min longer than basic reviews (unpublished data).

Collection of data

A three-step procedure was used to collect and classify DRPs. First, clinical pharmacists conducted the medication review and documented their work in a LIMM form. Five different pharmacists worked at the wards during the study period. Second, two student pharmacists classified the identified DRPs using a supporting form. This supporting form comprised a number of examples of DRPs for each type. The students were also informed about the classification procedure and they continually discussed any lack of clarity with a research pharmacist. Finally, two research pharmacists validated and, when necessary, reclassified the data in consensus.

Definition and classification of DRPs

A DRP was defined as “an event or circumstance that actually or potentially interferes with desired health outcomes”.² The DRP was classified as actual if an event was present (or if the patient’s health was currently affected) and potential if the event was not present but there was a risk of future events. The DRPs were then further classified into seven subcategories used by Cipolle et al ²⁰: need for additional therapy, unnecessary drug therapy, wrong drug (ie, ineffective drug, a drug therapy likely to be more effective exists, allergy or contraindication), dosage too low, adverse drug reaction, dosage too high and non-compliance.

Classification of actions taken to solve DRPs

The classification was done retrospectively by reviewing the LIMM medication review form and the health record. The actions taken were classified into 10 subcategories: withdrawal of drug therapy, change of drug therapy, initiation of drug therapy, decreased dose, increased dose, change in dosage interval, closer monitoring, information to the patient, change of another drug therapy (affecting the drug involved in the DRP) and change of drug formulation.

Statistics

Descriptive statistics are shown as medians (IQR), means (95% CIs) and frequencies or percentages when appropriate.

Ethics

The regional ethical review board in Lund, Sweden, did not consider ethical approval to be necessary and had no objections to the study. As data collection occurred within standard clinical care, routinely provided at the study site, patient consent was neither sought nor required.

Results

Description of the study sample

In total, 818 patients were eligible for inclusion in the study. Patients who did not receive a medication review (n=81) and patients for whom important demographic and study data were missing (n=18) were excluded. The final study population comprised 719 patients who received a medication review, of which 589 (82%) also received medication reconciliation at admission. The characteristics of the study population are summarised in table 1. Most of the patients (81%) were aged 75 or older. In median, patients were prescribed nine drugs for continuous use or as needed; 81% of the patients used five or more drugs, and 44% used ten or more drugs.

Table 1.

Characteristics of the study population

	Total population, n=719
Age, mean (SD), years	81 (10)
Sex, n (%) male	345 (48)
Housing before admission (%)
Home without homecare	403 (56)
Home with homecare	169 (24)
Nursing home	146 (20)
Length of stay, median days (IQR)
In the hospital	10 (6–16)
In the ward	8 (5–14)
No of drugs, median (IQR)
At admission
Continuous use	7 (5–11)
As needed	1 (0–2)
At discharge
Continuous use	8 (5–11)
As needed	2 (1–3)
During the hospital stay
Initiated	2 (1–4)
Stopped	1 (0–3)
Renal function reduction, n (%)*
No reduction	84 (11.7)
Mild, GFR 60–89 mL/min	157 (21.8)
Moderate, GFR 30–59 mL/min	313 (43.5)
Severe, GFR 15–29 mL/min	115 (16.0)
Kidney failure, GFR <15 mL/min	14 (1.9)
No information	36 (5.0)
Drug allergy or hypersensitivity documented in the EHR, n (%)	133 (18.5)

*Renal function was calculated using the Cockroft-Gault equation.

EHR, electronic health record; GFR, glomerular filtration rate.

Frequency of DRPs and actions taken

The clinical pharmacists conducted 292 basic and 427 extended medication reviews. A total of 5467 DRPs were identified in 695 patients (97%). Of these DRPs, 2164 (39.6%) in 573 patients (79.7%) were classified as actual DRPs (figure 1). The average number of DRPs in the total cohort was 7.6 (95% CI 7.2 to 8.0): 3.0 actual DRPs (95% CI 2.8 to 3.2) and 4.6 potential DRPs (95% CI 4.3 to 4.9).

Figure 1.

Illustration of the medication review process. The figure includes the number of identified drug-related problems (DRPs), the number of pharmacist-suggested adjustments or medication monitoring (described in the figure as adjustments in medication therapy), and the number of actual adjustments in medication therapy or medication monitoring. Data presented as number of actual DRPs (number of potential DRPs).

The actions of the pharmacists and physicians after identification of DRPs are summarised in figure 1. Overall, 80% (n=1740) of the actual DRPs and 8.3% (n=275) of the potential DRPs resulted in adjustment or monitoring of medication therapy, as a result of either the pharmacist’s suggestion or the physician’s initiative before the pharmacist’s suggestion.

The pharmacist suggested adjustment in drug therapy or monitoring of 1174 actual DRPs (54%), and 1037 of these treatment recommendations from the pharmacist (88%) were implemented by the physician. In 94 cases (8%), it was unknown whether the medication therapy was adjusted or monitored after the pharmacist’s recommendation. Either the actual DRP was no longer relevant for the patient or the physician never decided on an eventual adjustment.

Types of DRPs

The four most frequent types of actual DRPs were wrong drug, adverse drug reaction, need for additional therapy and unnecessary drug therapy (table 2). Actions to resolve the DRPs were most common for unnecessary drug therapy, adverse drug reactions and dosage too low (table 2 and figure 2). For adverse drug reactions, the physician initiated adjustments without recommendation from the pharmacist in just half of the cases. Inappropriate doses were most often adjusted after recommendation from the pharmacist.

Table 2.

Types of actual drug-related problems (DRPs) identified during the medication reviews (both basic and extended)

Actual drug-related problems	Number of patients with DRPs (prevalence) n=719	Number of DRPs identified for each type (percentage of total number of DRPs) n=2164	Number of DRPs acted on (percentage of number of DRPs identified)
Wrong drug	291 (40%)	601 (28%)	431 (72%; 431/601)
Adverse drug reaction	212 (29%)	429 (20%)	373 (87%; 373/429)
Need for additional therapy	275 (38%)	397 (18%)	301 (70%; 301/397)
Unnecessary drug therapy	211 (29%)	360 (17%)	353 (98%; 353/360)
Dosage too high	165 (23%)	239 (11%)	173 (72%; 173/239)
Dosage too low	93 (13%)	129(6.0%)	106 (82%; 106/129)
Non-compliance	6 (0.8%)	9 (0.4%)	3 (33%; 3/9)
All	573 (80%)	2164 (100%)	1740 (80%; 1740/2164)

Data presented as number (percentage).

Figure 2.

Percentage of actual drug-related problems (DRPs) acted on without recommendation from the pharmacist or after recommendation from the pharmacist, respectively, and percentage not acted on by the physician for each type of DRP. No action taken=no recommendation made by the pharmacist and no action taken by the physician.

The three most common types of adjustments made were withdrawal of drug therapy, change of drug therapy and initiation of drug therapy (table 3).

Table 3.

Adjustment of medication therapy for actual drug-related problems

Adjustment made	Number of adjustments in total (number of adjustments as a result of the pharmacist suggestion)
Withdrawal of drug therapy	810 (359)
Change of drug therapy	327 (269)
Initiation of drug therapy	285 (188)
Decreased dose	121 (80)
Increased dose	70 (46)
Change in dosage interval	61 (53)
Closer monitoring	39 (30)
Information to the patient	14 (14)
Change of another drug therapy	12 (12)
Change of drug formulation	1 (1)
Total	1740 (1052)

The most frequent medication groups involved in DRPs on ATC first level were drugs for the cardiovascular system (26%) and the nervous system (25%) and on ATC second level were analgesics and antithrombotics (figure 3). Altogether, the 10 most common medication groups accounted for 58% of the identified actual DRPs. In the cases where untreated or not optimal treated indication was identified and additional therapy was suggested, the most frequent symptoms or diseases were pain (15%), heart failure (9.1%) and osteoporosis (7.6%).

Figure 3.

Ten most frequent medication groups involved in actual drug-related problems (DRPs) and types of DRPs within each medication group. Missing data for four DRPs. RAAS, renin-angiotensin-aldosterone system.

Warfarin, digoxin and furosemide were the medications most frequently involved in the actual DRPs. Drugs and type of DRPs in percentages for each medication are shown in figure 4. The 10 most common medications accounted for 27% of the total number of actual DRPs. Too high doses were more common in drugs with a narrow therapeutic window (ie, warfarin and digoxin). Unnecessary drug (no indication for treatment) was most common for ipratropium and tramadol. Spironolactone was the drug with the highest proportion of adverse drug reaction, mostly hyperkalaemia.

Figure 4.

Ten most frequent medications involved in actual drug-related problems (DRPs) and types of DRPs for each medication. Missing data for four DRPs. ASA, acetylsalicylic acid.

Discussion

The LIMM model has been developed during more than a decade, have shown clinical, economical and humanistic outcomes, and has been summarised elsewhere.²¹ The total project now includes more than 20 publications and is the background for four PhD theses. The model has received several awards and is the basis for introducing a full clinical service implemented in all hospitals in the south of Sweden, spreading also nationally and internationally.

This study is the final publication of the model focusing on the medication reviews and the process of identification and resolution of DRPs. It describes a team-based intervention including a clinical pharmacist that aimed to individualise and optimise drug treatment for inpatients. In the included population of elderly patients with polypharmacy, 80% had at least one actual DRP, with an average number of three DRPs per patient. Of these problems related to drug therapy, 80% were acted on by the physicians during hospital stay thus providing external validity and effectivity of the model. In a previous study in the same population, the clinical significance was assessed in a random sample of 132 patients.²² A total of 197 recommendations were assessed, 178 (90%) of these were implemented and most of these recommendations, 170 (83%), were ranked three or higher on a six-point scale for clinical significance (ie, at least somewhat significant). Other hospital-based studies, predominantly involving poly-medicated older patients, have reported similar or slightly lower numbers of DRPs per patient^{9 11 23} and similar proportions of patients experiencing at least one DRP (ie, 75%–85%).^{11 23} By minimising prescribing errors and DRPs, medication reviews might improve patient outcomes. A recent systematic review and meta-analysis²⁴ and a MiniReview²⁵ demonstrated that regular ward-based pharmacist input may improve quality and safety of medication use, although the evidence was of very low to moderate quality due to risk of bias, imprecision and inconsistency. There are evidence of benefits of activities both at admission, during hospital stay and at discharge,²⁴ but it seems like multifaceted pharmacist interventions are more effective than interventions with a single component.²⁵ In a recent Danish randomised controlled study including 1467 patients, an extensive multifaceted clinical pharmacist intervention, with follow-up after discharge, showed significant and clinical important reductions in total hospital readmissions.²⁶ The optimal level of involvement of ward-based pharmacists needs further research.^{24 25}

Wrong drug and adverse drug reaction were the most common DRPs. Another Swedish study using the same classification system, which was performed in a similar setting but with slightly older patients, reported a similar pattern of DRP types.⁹ However, there was one significant exception though; the DRP class “wrong drug” was the most common in our study but the least common in the study by Gillespie et al, most likely due to the different interpretation concerning the definition of DRPs, which highlights the difficulties in classifying DRPs in a consistent way across studies. The most common adjustment made in drug therapy was withdrawal of a medication, which is in line with other studies of medication reviews in elderly hospitalised patients.⁹ Even if the most common intervention after medication review was discontinuation of a drug, the median number of drugs increased during hospital stay. This may reflect initiation of appropriate drug therapy during hospital stay, often time restricted treatments like antibiotics or pain medications. Nevertheless, it highlights the need for continuous evaluation of drug therapy, especially after hospitalisation, to decrease the burden of polypharmacy in the elderly population.

The finding that physicians to a high degree identify adverse drug reactions reflects that such DRPs can be detected when assessing symptoms and diagnosis. On the other hand, the need for changing drug or adjusting doses due to, for example, contraindications or impaired kidney function was most often initiated by the pharmacist. Such interventions are more likely to be identified after a thorough clinical medication review, and a clinical pharmacist can contribute to the medical team by performing such reviews.

Commonly used medication groups, in particular analgesics, antithrombotics, phycholeptics, diuretics and antibacterial drugs, were likely to be involved in the DRPs. These findings are consistent with other studies.²⁷ Thus, clinical pharmacists and physicians should pay particular attention to some substances and type of DRPs that are usually involved.

The description of the medication review process and the characteristics of the identified DRPs are important to study in order to understand and optimise the effectiveness of the intervention and the potential clinical outcomes. In this study, we more specifically also studied medication groups and substances frequently involved in DRPs in relation to DRP type. One strength of this study was the thorough analysis of the medication review process, from the identification of DRPs to the actions taken for each type of DRP. Another strength was the size of the study, given that the study is not based on population databases. Furthermore, the prospective design also minimised the risk for overlooking DRPs and for registering clinically non-relevant DRPs that in fact are appropriate for a specific patient.

There are also some limitations associated with our study. For example, only five pharmacists were involved and the intervention was performed on two internal medicine wards in one university hospital, which may limit the generalisability of the results. The fact that the clinical pharmacist conducted medication reviews according to a strict protocol and the fact that many physicians and nurses participated compensate to some extent. Another limitation was that the clinical relevance of identified DRPs has not been assessed for all patients. However, for a random sample of these patients, the clinical relevance has previously been reported.²²

We conclude that DRPs are common among elderly patients admitted to hospital, mainly wrong choice of drug and adverse drug reactions. The most common intervention for solving a DRP was to withdraw the drug in question, reflecting the risk for inappropriate polypharmacy in the older population. A limited number of drugs constitute a substantial proportion of DRPs. Hence, systematic identification of high-risk medications and common DRP types enables targeting of prioritised patients for medication reviews.

Key messages.

What is already known on this subject

• Drug-related problems (DRPs) have been shown to cause admission to hospital and a considerable number of these are preventable.

• Approaches to identify and decrease DRPs are therefore of great interest.

• The Lund Integrated Medicines Management (LIMM) model is a systematic hospital-based clinical pharmacy service developed and researched during more than a decade.

What this study adds

• In real-life practice, the LIMM model identifies an average of three DRPs per patient; 573 patients (80%) had at least one actual DRP.

• When the pharmacist suggested an adjustment, the physician implemented 88% (1037/1174) of the recommendations.

Data availability statement

Data are available on reasonable request. Relevant patient data were collected manually in a form and used for clinical purposes. After discharge, a code number was inserted and patient names deleted. This code was used when processing data.

Ethics statements

Patient consent for publication

Not required.