Contribution of drug related problems to hospital admission in the elderly

Abstract

Purpose

To investigate the frequency and type of drug related problems (DRPs) in geriatric patients (> 65 years), and to assess their contribution to hospital admission; to explore the appropriateness of drug therapy according to the Beers' criteria.

Methods

Cross-sectional observational survey of 110 elderly patients admitted during three non-consecutive months to the geriatric ward of a university hospital. Explorative assessment of appropriateness of drug therapy prior to hospital admission based on the Beers' criteria

Results

A DRP was the dominant reason for hospital admission in 14 out of the 110 patients (12.7%); for another 9 patients (8.2%), a DRP was partly contributing to hospital admission. For these 23 patients, adverse drug reactions and noncompliance were the most important types of DRPs. We found no relationship between drug related hospital admission and intake of a drug listed in the Beers criteria for inappropriate drug use in the elderly. Patients admitted for a DRP took more drugs before admission than patients admitted because of other reasons.

Conclusions

DRPs are an important cause for admission on the geriatric ward of our hospital. The drugs causing DRPs in this study were not those listed in the Beers list of inappropriate drugs in the elderly.

Introduction

Numerous studies show that elderly patients often experience drug related problems leading to hospital admission. The reported percentages vary however considerably, from 4 to 30% (1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13). The majority of these problems concerns adverse drug reactions, and several studies have estimated that 50% to 97% of drug related problems in the elderly were avoidable (1, 2, 4, 5, 6, 14).

The aim of this study was to investigate in depth the typology of drug related problems (DRPs) in geriatric patients (> 65 years) admitted to the acute geriatric ward of a university hospital, and to assess the contribution of these DRPs to hospital admission. In addition, we wanted to evaluate the appropriateness of drug therapy in these patients, using the Beers’ criteria of inappropriate prescribing (15, 16).

Methods

Setting and study design

This study was performed on the geriatric ward of the Ghent University Hospital, Belgium (30 beds), after approval of the institutional Ethics Committee. More than 75% of the elderly persons admitted to our institution enter the hospital via the emergency department. If no surgery is planned, they are transferred in function of their medical condition to one of the wards for internal medicine; in case they do not present a specific condition, or if they are confused or suffer from dementia, they are transferred to the geriatric ward. Patients hospitalised at the geriatric ward are on average older than other hospitalised patients; other wards were older persons are cared for include neurology, cardiology and pneumology.

The patients admitted during 3 non-consecutive periods (October 2002, March 2003 and June 2003) were screened at admission for drug related problems. Drug use prior to and during admission was registered, and length of stay and mortality during hospitalisation was recorded. A panel consisting of the treating geriatrician, a clinical pharmacologist and a hospital pharmacist reviewed the clinical chart and the drug list of all admitted patients.

Data collection

Drug use prior to admission was registered as usual. After initial registration in the emergency department, this is further completed at the geriatric ward, where both nurse and physician question the patient and/or his/her family. If there is no referral letter, or when there are doubts about the home medication, the general practitioner or the nursing home is contacted. Special attention is paid for drugs started less than one month before hospital admission, and for drugs in relation to the central nervous system (e.g. hypnotics). Drug use during the hospital admission was found in the nursing file, and was double checked with the pharmacy files for concordance. Drugs at discharge were registered by means of the discharge letter.

All drugs taken prior to admission, during and after admission, were introduced in an Access database, based on a list of drugs available in Belgium.

Definition of drug related problems

The panel determined whether no, one or several DRPs were present in each patient, and classified each DRP, using the typology of Hallas (17).

In this classification (see Table 1), a distinction is made between adverse drug reactions (ADRs) and drug therapy failures (DTFs). An adverse drug reaction is defined according to the WHO definition (18) and further classified as occurring with normal use, overdose, contra-indication or drug interaction. A drug therapy failure is defined as a lack of therapeutic effect due to either drug non-compliance, a too low prescribed dose, or a drug-drug interaction. In this classification, untreated indication and drug use without (validated) indication are not taken into consideration.

Table 1.

Assessment of drug related problems according to Hallas (6)

Typology of drug-related problems Adverse drug reactions (ADRs) occurring during normal use caused by overdose caused by non-respect of a contra-indication caused by drug interaction Drug therapy failures (DTFs) caused by drug non-compliance caused by underdosing caused by a drug-drug interaction Criteria for assessment of causality of ADRs 1. a known ADR or toxic reaction; 2. a reasonable temporal relationship between beginning drug therapy and onset of ADR; 3. disappearance of the ADR on discontinuation or dosage reduction; 4. inability to explain the symptom or event by other conditions or predisposition of the patient; 5. reappearance of the symptoms on re-exposure, or laboratory tests showing toxic drug levels or drug-induced metabolic disturbances that explain the symptom; The causality of the relationship is accepted as ‘definite’ when all 5 criteria are satisfied, ‘probable’ when 1 through 4 are satisfied, ‘possible’ when 1 through 3 are satisfied, ‘unlikely’ or ‘unevaluable’ when the necessary information cannot be obtained or other conditions are considered to be more likely the cause of the symptoms. Criteria for the assessment of causality of Drug Therapy Failures 1. the symptoms of the disease are known to reappear at insufficient doses; 2. the symptoms are not likely to have been caused by progression of the disease; 3. a reasonable temporal relationship between the start of inadequate dosage and the appearance of symptoms; 4. resolution of the symptoms upon adjustment to an adequate dose; 5. no other conditions that explain the symptoms; 6. drug levels clearly below the therapeutic range or clear evidence of intake of an insufficient dose. The causality of the relationship is accepted as ‘definite’ when all criteria are satisfied, ‘probable’ when 1 through 5 are satisfied, ‘possible’ when 1 through 4 are satisfied, and ‘unlikely’ or ‘unevaluable’ when the necessary information cannot be obtained or other conditions are considered to be more likely the cause of the symptoms. Contribution of ADR or DTF to hospital admission - ‘dominant’: drug-related symptoms are the main reason for admission and no other symptoms contribute significantly - ‘partly contributing’: drug-related symptoms are a substantial reason for admission, but other factors are also present - ‘less important’: drug-related symptoms play a minor or uncertain role, and the patient would most likely have been admitted without them - ‘not contributing’: symptoms other than the drug-related symptoms are the main reason for admission

To confirm the drug related nature of the problem, in a next step, the causality of the drug related problems is evaluated with a set of six criteria for ADRs, and a set of five criteria for DTFs (see Table 1). In a final step the potential contribution of a DRP to the hospital admission is assessed by classifying the DRP as a dominant, a partly contributing, a less important, or a not contributing factor to the hospital admission.

ADRs with unlikely or unevaluable causality and DTFs with a possible, unlikely or unevaluable causality were not taken into account for analysis. Furthermore, an admission was considered as related to an ADR or DTF when the symptoms were found to be the dominantly or partly contributing reason for the hospital admission. All this was assessed by the panel, after discussion until consensus was reached.

Criteria for inappropriate drug therapy

The criteria developed by Beers and colleagues, updated in 1997, were used to assess inappropriate drug use (16). These criteria list drugs that should generally be avoided in the elderly, doses or frequencies of administration that should not be exceeded, and drugs that should be avoided based on drug-disease interactions. We only used the list of drugs that should generally be avoided independent of diagnosis, and translated this to the drugs available on the Belgian market (list available upon request).

Data analysis

Patients were divided into two groups according to whether or not a drug related problem, dominantly or partly contributing, was present at admission. We then tested statistically for various variables the difference between the two study groups, in order to detect risk factors for drug related hospital admission.

The Mann-Whitney U-test was used for assessing differences in age, length of stay, total number of drugs and number of inappropriate drugs taken before admission as well as at discharge. We also evaluated whether gender, residence in a nursing home, previous admission during the last six months, or death during admission was related to a DRP that dominantly or partly contributed to hospital admission by using the Fisher’s exact test. Furthermore, the overall difference between the number of drugs taken at home and the number of drugs taken at discharge was tested by the Wilcoxon signed-ranks test.

Statistical analysis was performed using the R Development Core Team software (version 2.6.1, 2007) (19).

Results

During the three non-consecutive months of the study, 110 patients (41 men, 69 women) were admitted to the geriatric ward, directly, via the emergency department or after transfer from another hospital ward. Their mean age was 82.7 years and their mean length of stay at the geriatric ward was 16.6 days (1 – 50, median 14 days). Twenty percent of the patients (n=22) came from a nursing home, and thirty percent (n=33) had already been admitted to our hospital during the previous six months. Ten patients died during the hospital stay. The mean number of drugs per patient at admission was 5.9 (0 – 15, median 6) (Table 2). Overall, the number of drugs at discharge was significantly increased (p < 0.001), with a mean of 7.1 drugs per patient (0 – 15, median 7). At discharge, there was a decrease in the use of psychotropic drugs, and an increase in the use of laxatives.

Table 2.

Demographic data of the patients admitted during the study period (n=110)

	Mean	Median	Range	SD
Age (years)	83.2	82.7	65.5 – 102.5	7.5
Total length of stay (days)	19.6	16	1 – 85	14.3
Length of stay at the geriatric ward (days)	16.6	14	1 – 56	10.9
Number of drugs taken before admission (n=100)	5.9	6	0 – 15	3.1
Number of drugs taken at discharge (n=100)	7.1	7	0 – 15	3

Drug related problems

The panel (treating geriatrician, clinical pharmacologist and hospital pharmacist) identified the presence of a drug related problem with a definite, probable or possible causality at the time of admission in 35 patients (31.8%), namely 22 ADRs and 13 DTFs. For 31 of the 35 DRPs, there was a probable causal relationship between the drug and the symptoms, for 3 a definite relationship and for one DRP the causality was estimated as possible. For 14 of the 110 patients (12.7%), the DRP was considered to be the dominant reason for hospital admission; for another 9 patients (8.2%), the DRP was thought to partly contribute to hospital admission. Seven DRPs were found to be less contributing to hospital admission, and five DRPs were found not to contribute. We focused on the problems that contributed dominantly or partly to hospital admission (n=23).

Drug related hospital admissions

As shown in Table 3, the DRPs that were dominantly or partly contributing to hospital admission concerned 14 ADRs (8 normal use, 4 intentional overdose, one unintentional overdose and one drug interaction), and 9 DTFs (6 non-compliance and 3 subtherapeutic dose). The DRPs concerned central nervous system medication (9), antidiabetics (3), respiratory drugs (3), cardiovascular drugs (2), anti-inflammatory drugs (2), and four other classes. Nineteen drugs were taken orally, 3 by inhalation (respiratory drugs) and 1 drug was administered subcutaneously (insulin).

Table 3.

Drug related problems that contributed dominantly or partly to hospital admission (n=23)

Type of DRP	Drug	Symptoms	Causality	Contribution to admission
ADR drug interaction	risperidone,	muscle weakness, dehydration, disorientation	probable	partly contributing
(1)	citalopram, mirtazapine
ADR normal use	amiodarone	toxic hepatitis	probable	dominant
(7)	diclofenac	stomach irritation, abdominal pain	probable	partly contributing
	metformin	symptomatic hypoglycaemia	probable	dominant
	methylprednisolone	haemoptysis	probable	partly contributing
	naproxen	gastric haemorrhage	probable	dominant
	salbutamol	supraventricular tachycardia	probable	partly contributing
	thalidomide	somnolence	probable	partly contributing
ADR unintentional	lorazepam	fall	probable	dominant
overdose (2)	metformin	symptomatic hypoglycaemia	probable	dominant
ADR intentional overdose	acetylsalicylic acid	poisoning (established)	probable	dominant
(4)	diazepam	somnolence, confusion, falls	probable	partly contributing
	opioids	somnolence, intoxication	probable	dominant
	tilidine/naloxon	somnolence, ataxia, nausea	probable	partly contributing
DTF non-compliance	amoxyclav	dyspnoea, coughing, sputa	definite	dominant
(6)	anti-asthmatics	COPD exacerbation, dyspnoea	probable	dominant
	antidepressive agents	depressive, walking around at night	probable	dominant
	formoterol + corticoid	dyspnoea, coughing	definite	dominant
	levothyroxine	dyspnoea, stridor	probable	dominant
	neuroleptics	confused, aggressive	definite	dominant
	DTF subtherapeutic dose (3) anti-epileptics	seizure with absence and urine loss	probable	partly contributing
	insulin	symptomatic hyperglycaemia	probable	dominant
	sotalol	atrial fibrillation	probable	partly contributing

ADR = adverse drug reaction; DTF = drug therapy failure

Patients admitted dominantly or partly because of a drug related problem, took a significantly higher number of drugs before admission (p=0.03), and at discharge (p=0.03). Furthermore, patients who had been hospitalised during the previous six months, were more likely than the others to develop a DRP leading to hospital admission (p=0.04).

No relationship between drug related hospital admission and sex or coming from a nursing home, was found. The patients who were admitted because of a DRP were slightly younger (mean 80.3 years) than those admitted for other reasons (mean 83.9 years) (p=0.04). The patients admitted because of a DRP stayed longer in the hospital in comparison with the others; however, this difference (4,5 days in total lenght of stay) was not statistically significant. No relationship was found with death during hospital stay (see Table 4).

Table 4.

Drug related hospital admissions (DRHA’s) in relation to demographic data and other variables

	Mean	DRHA (n=23)	Non-DRHA (n=87)	p-value
Age (years)	83.2	80.3	83.9	0.0436
Total length of stay (days)	19.6	23.2	18.7	0.248
Length of stay at the geriatric ward (days)	16.6	19.4	15.9	0.395
Number of drugs taken before admission	5.9	6.9	5.7	0.0285
Number of drugs taken at discharge	7.1	8.3	6.8	0.0313
	Number of patients	DRHA (n=23)	Non-DRHA (n=87)	p-value
Readmission	33	11	22	0.0437
Death during hospital stay	10	0	10	0.117
Coming from a nursing home	22	4	18	1.0
Taking an ‘inappropriate’ drug	22	6	16	0.534
listed by Beers, at admission

Inappropriate drug use according to Beers’ criteria

Before hospital admission, 22 patients (20%) took a drug listed in the Beers’ criteria of inappropriate drugs (9 different drugs); this concerned 5% of the overall drug use before admission. This number was reduced at discharge, to 18 patients (16%) taking an ‘inappropriate’ drug (8 different drugs), concerning 3% of the overall drug use at discharge. Two out of the 23 drugs that caused dominantly or partly hospital admission were ‘inappropriate’, namely lorazepam (> 2.5 mg) and diazepam. No relationship was found between the intake of an ‘inappropriate’ drug listed by Beers, and the occurrence of a DRP leading dominantly or partly to hospital admission (p>0.05).

Discussion

Incidence and types of drug related hospital admissions

We found that drug related problems at the time of hospital admission were common, and one out of five patients was suffering from a drug related problem that required hospitalisation or contributed to it. When comparing studies about this subject, we have to consider the definition of drug related problems, and the definition of drug related hospital admissions used in these studies.

The definition of drug related problems varies from study to study. In all the publications in the reference list below, adverse drug reactions were considered, although not all studies included the ADRs originating from intentional overdose. In some studies also inappropriate drug choice, untreated indications and drug use without indication were taken into account. In our study, we did not include these aspects, since it was our goal to focus on problems that could be clearly linked to the use of drugs, and since information about previous medical decisions would probably not often be available.

Applying the method established by Hallas et al. (17), we found an incidence of 20.9% of drug related hospital admissions. When we limit to ADR related hospital admissions, we found an incidence of 12.7% and when excluding intentional overdoses, the incidence was 9.1%. In other studies, similar incidences of ADR related hospital admissions were found. Beijer et al. (1) published a meta-analysis in 2002, in which 17 studies about ADR related hospitalisations in the elderly were included, and found a mean incidence of 16.6%. The authors of this meta-analysis remarked that the studies included were relatively small sized, varying from 100 to 1988 hospitalisations, and that large studies (>2000 patients) were lacking. Later, larger studies about drug related hospitalisations of the elderly were performed in Great Britain (11) (18000 patients), Italy (12) (12000 patients), and in the Netherlands (13) (> 660000 patients). The incidence of ADR related hospital admissions (dominant reason for hospital admission) was estimated at respectively 5.2% (11), 3.4% (12), and 3.2% (13). Only a few publications also included other types of drug related problems, such as non-compliance, subtherapeutic dose, and inadequate monitoring. A study performed in the UK in 1997 with about 1.000 elderly patients found an incidence of 5.3% overall drug related hospital admissions (5). Another study in the UK in 2001 with more than 4.000 patients of all ages, found 6.5% of admissions to be drug related (7). As discussed in the paper of Beijer et al. (1), it seems that larger studies seem to find lower incidences of drug related admissions. This might be due to a less intensive or retrospective screening system, while smaller sized studies allow a more accurate recording of drugs and symptoms based on medical, nursing and laboratory findings, and a multidisciplinary assessment of causality and contribution to hospital admission. Another factor which could explain the relatively high incidence of drug related hospital admissions we found is the fact that our study was performed at a geriatric ward, where concern about medication use is high, and registration and evaluation of drugs used before admission is performed extensively. Furthermore, the patients we studied took a high number of drugs, which is a well known risk factor for developing adverse drug reactions (20, 21, 22, 23), as confirmed in our study. Our finding that the number of drugs at discharge was higher in the patients hospitalised dominantly or partly because of a drug related problem, requires careful evaluation, since hospitalisation is most often a milestone in worsening clinical condition, including step-ups in pharmacotherapy. In our patients, drug related hospital admissions were also more frequent in somewhat younger patients. Although statistically significant, we have no clinically relevant interpretation for it.

Criteria for inappropriate use

Only a small percentage of the drugs used before admission and at discharge were inappropriate according to the Beers’ criteria. We found no relationship between intake of ‘inappropriate’ drugs, and the occurrence of a DRP leading dominantly or partly to hospital admission. Beers’ criteria have been criticized, since they do not identify all causes of potentially inappropriate prescribing (24). Indeed the process of selecting a drug in relation to indication and contra-indications, choice of dosage and duration of therapy, and the monitoring for adverse drug reactions or interactions, is more important than the drug itself. It has therefore been argued that although the Beers’ criteria have been widely used and represent a standardized tool for pharmacological research, they cannot be considered as a substitute for careful clinical judgement (25).

Strengths and weaknesses

The strength of this study is that drugs used before, during and after admission, were recorded accurately, and the assessment of the contribution of drug related problems to hospital admission was performed through a process of multidisciplinary consensus reaching between the panel members. Therefore, we choose for a small but in-depth study. Secondly, we tried to define relevant drug related problems in such a way that they could be clearly linked to the use of drugs. This led us further into understanding the nature of drug related problems in the elderly, i.e. the sensitivity of frail elderly people to well-known adverse drug reactions and the risk of non-compliance with deterioration of diseases as a consequence. This means that drug use in the elderly is a difficult equilibrium, with small changes in response resulting in adverse drug reactions or worsening of the clinical condition; therefore, follow-up and monitoring is extremely important to reach or maintain this equilibrium.

We are aware that our study has several limitations. A formal screening method for identification of drug related problems such as the Medication Appropriateness Index (MAI) was not used (26, 27), nor a method for inter-rater variability scoring. Furthermore, the number of patients included was relatively small.

Prevention strategies

Inappropriate prescribing for frail elderly persons as well as inadequate communication between primary and hospital care have been mentioned in the literature as causes of drug related problems (14, 28, 29, 30, 31, 32). Appropriate prescribing is not only about drug choice, but includes also careful evaluation of doses, duration of therapy, monitoring for adverse reactions and drug-drug interactions. As elderly people are often treated by several physicians, there is a risk for polypharmacy and therefore the occurrence of drug related problems. A number of actions can be taken in hospitals, to stimulate appropriate prescribing and to assure adequate communication between primary and hospital care:

• -education of caregivers;
• -accurate recording of drugs used;
• -multidisciplinary medication review with advice of clinical pharmacists and clinical pharmacologists;
• -informing patients about changes in drug regimens and about newly started drugs;
• -informing first-line caregivers (general practitioners, care workers) about changes in drug regimens and advice for follow-up;
• -electronic patient files and computer assisted prescribing.

For these actions, attention should focus mainly to patients taking a high number of drugs.

Conclusion

This study confirms that drug related problems represent an important problem in geriatric patients admitted to the hospital, and that there is a significant contribution of these problems to hospitalisation (20.9%). We found a positive correlation of these problems with the number of drugs used, but not with treatment with inappropriate medications defined by Beers. Although the study size and the methodology used can be criticised, we consider this study to be useful because the analysis was thoroughly performed. In order to prevent drug related problems in the elderly, we recommend to focus on multidisciplinary medication review for elderly patients admitted in hospitals, and to pay attention to all aspects of pharmacotherapy, e.g. side effects, under- or overdose, drug-drug interactions, duration of therapy and non compliance.

Acknowledgements: The authors are indebted to the participating geriatricians prof. dr. M. Afschrift, prof. dr. N. Van den Noortgate, dr. A. Velghe, dr. H. Martens, dr. N. Van Doninck, and to Head Nurse H. Van Doninck, and the patients of the geriatric ward. This study was not funded and the authors have no conflicts of interest.

Financial disclosure: None of the authors had any financial interest or support for this paper.