Introduction
A novel approach was used in this study, to assess whether human error theory can be used to explain the reasons behind non-adherence. A better understanding of whether non-adherence can be considered within the human error framework may mean that appropriate interventions to maintain adherence can be initiated at an earlier stage. The aim of this study was to investigate whether human error theories can be applied to non-adherence in medication taking by those on long-term medication.

Methods
This study was conducted on a sample of 105 patients. Patients were recruited from illness groups where the burden of disease is high, where there are particular compliance issues and where non-adherence may have a serious impact on outcome.

A multi-method approach was adopted, contacting patients by telephone and conducting semi-structured interviews as well as sending out postal questionnaires. This combination of methods was chosen to maximise the relative benefits of each of these questionnaire styles. Where patients reported non-adherence (missing doses of medication) in the telephone interview, the reasons for this were explored.

Results
From a sample of 105 patients, 87 were successfully interviewed (83% response rate). From the 87 patients interviewed, 74 agreed to complete the postal questionnaire with 58 returned (78% response rate). There were 42 patients from the 87 interviewed reporting non-adherence. Of these, 17 were non-adherent within the 7 days prior to interview.

The majority of cases of non-adherence were classified as skill-based errors, where patients simply forgot to take their medicines for a variety of reasons. There were some examples of rule-based errors, the application of a 'bad' rule or the misapplication of a 'good' rule. There were no reports of knowledge-based errors, with violations occurring primarily in patients who were taking diuretics and who chose not to take them before travelling or going out for the day. Performance influencing factors were also categorised, as were latent failures that may have led to the error. Examples of each of these will be given.

Conclusions
The human error approach is a novel approach in assessing non-adherence, and focuses on the system, or environment, in which patients take medicines rather than on the individual. It is useful in highlighting deficiencies in the system that may lead to non-adherence. The human error theory is useful in its application to non-adherence but is limited in its use due to requiring complete factual data of all the events leading to an episode of non-adherence.