Summary

Background: Preventable adverse drug events are caused by errors in the medication use

Preventable adverse drug events are caused by errors in the medication use

process, and are of particular interest when designing interventions to improve the quality

of medication therapy. Type 2 diabetes became the case because a large proportion of

patients are undertreated and not monitored as recommended; yet, the epidemiology of

preventable adverse drug events in type 2 diabetes is largely unknown. The aims of the

studies were to develop quality indicators for preventable adverse drug events in patients

with type 2 diabetes, and apply a new register-based model with the ability to assess the

epidemiology and economic impact of preventable adverse drug events in patients with

type 2 diabetes, in order to give health care decision makers a clinical and health economic

rationale for prioritizing interventions against medication errors.

Methods: A systematic literature review of the available literature on preventable adverse

A systematic literature review of the available literature on preventable adverse

drug events was conducted to describe the incidence and characteristics of preventable

adverse drug events in the primary care setting (Article 1).

Construction of the preventable adverse drug events assessment model required four steps.

The first step consisted of the development, validation, and feasibility of preventable

adverse drug event indicators for type 2 diabetes (Articles 2 and 3). Preventable adverse

drug event scenarios were identified through a literature review. Face, content and construct

validity was assessed using a Delphi consensus study. The feasibility of a computerised

search for indicator positives within Danish health care registers was determined by

assessing the operationalisability of the indicators, and the availability of the required data.

The second step in the model consisted of the identification of preventable adverse drug

events. To identify the most frequent preventable adverse drug events, a database

containing health care data for a random 10% of the Danish population receiving oral

antihyperglycaemic treatment (9,791 persons) was set up, and indicator positives identified

(Articles no. 2 and 3). The third step in the model consisted of a risk assessment of

preventable adverse drug events, including an assessment of clinical areas that need quality

improvement. One study concerned the four indicators on inadequate preventive treatment

after acute myocardial infarction (AMI) (Article 4), and another study concerned the four

indicators on inadequate monitoring of HbA1c resulting in admission with hyper- or

iii

hypoglycemia (Article 5). In the AMI study, a register-based study of the total Danish

population (8,275 persons) receiving oral antihyperglycaemic treatment and surviving an

AMI was conducted. The study period was 1 January 2001 to 31 December 2004. Patients

were followed until their first recurrent AMI. The proportion of patients in inadequate

treatment, and the risk of recurrent AMI associated with inadequate treatment were

determined. Statistical analysis was performed using a Cox regression model and a casecrossover

design. In the HbA1c study, a register-based study of the total population in the

Danish county of Århus receiving oral antihyperglycaemic treatment (18,922 persons) was

conducted. The study period was 1 January 2000 to 31 December 2004. The rate of

indicator positives; the status of HbA1c monitoring; stratification between treatment

settings; and high risk patient groups were assessed. The risk of diabetes-related hospital

admission associated with HbA1c monitoring and treatment was determined using logistic

regression.

The fourth step in the model was a health economic evaluation of the cost-effectiveness of

shifting patients from inadequate to adequate medical treatment. The database used for the

AMI indicator study formed the database for this study. Five post AMI treatment scenarios

were analysed, and incremental cost-effectiveness ratios calculated.

Results: The systematic literature review (Article 1) revealed that preventable adverse drug

The systematic literature review (Article 1) revealed that preventable adverse drug

events are common in primary care, with many resulting in hospitalisation. Projected to a

Danish setting, 216,000 patients will experience a preventable adverse drug event each

year, and 162,000 of these patients will need hospital admission. Quality improvement

interventions should target errors in prescribing and monitoring, especially for patients

using cardiovascular drugs, analgesics and antihyperglycaemic agents.

The indicator development study resulted in 27 preliminary preventable adverse drug

events indicators for type 2 diabetes, of which 18 (67%) achieved consensus (Article 2). All

indicators were operationalisable, but incomplete information on laboratory data and overthe-

counter medicine caused inaccurate incidence estimates. Therefore, the two risk

assessment studies relied on indicators where laboratory data were not required, or relied on

a diabetes specific database containing the necessary laboratory data.

iv

The screening study (Article 3) revealed that most cases of preventable adverse drug events

concern admissions with hyperglycemia, reduced renal function and recurrent AMI, related

to inadequate treatment or monitoring.

The AMI indicator study (Article 4) revealed that 77% of type 2 diabetes patients do not

receive recommended treatment after an AMI and 19% experience a recurrent AMI after a

median follow-up time of only 1.75 years. Inadequate preventive treatment increases the

risk of recurrent AMI up to five times. Even though underdosed or absent antiplatelet

treatment doubles the risk of recurrent AMI, 29% do not receive such treatment. Also, the

risk of recurrent AMI decreases as the number of recommended medications or the doses

increase; yet, beta-blockers and lipid-lowering agents were underdosed.

The economic evaluation based on the AMI study (Article 6) revealed, that from a public

health care systems' point of view, providing intensive cardioprotective treatment

according to already accepted guidelines to type 2 diabetes patients is cost-effective.

The HbA1c study demonstrated how diabetes-related hospital admissions are frequent, and

how preventable adverse drug events are particularly frequent in insulin users. Important

target groups for future quality improvement initiatives are the small group of unmonitored

patients (3.3 %); the 20 % of the population being hyperglycaemic, particularly those who

do not receive antihyperglycaemic treatment; patients receiving cardiovascular but no

antihyperglycaemic treatment.

Implications: The developed model for assessment of the epidemiology of preventable

The developed model for assessment of the epidemiology of preventable

adverse drug events may be used at the system level, health care practitioner level, and

patient level. At the system level, the model may assist the tailoring of interventions aimed

at preventing the most frequent medication errors and such which may have serious

consequences, and eventually assist in restructuring the medication use system to achieve a

system less sensitive to errors and safer for patients.

At the health care practitioner level, patient databases may be screened for indicator

positives to provide an assessment of the quality of provided health care. At the patient

level, the indicators may be used to identify medication errors in order to prospectively

avoid the manifestation of adverse outcomes.