We found obvious deficits in care regarding guideline adherent drug therapy for hypertension, diabetes mellitus type two, heart failure, atrial fibrillation and secondary prevention in cardiovascular diseases. About a sixth of all quality indicators in our study were not fulfilled according to current guideline recommendations. In more than half of these quality indicators the patients did not know why they were not prescribed a particular drug, thus making us look at the physician as the one responsible for non-adherence. The most frequent reason for physicians to deviate from guideline recommendations was that they falsely assumed that a certain prescription was not indicated or necessary.
There are several possible explanations for the fact that the treatment of patients is not always consistent with evidence based recommendations. According to our study, the most important cause appears to be the physician not providing a particular treatment. This may be due to physicians’ lack of awareness regarding the existence of a guideline, or lack of familiarity with a guideline, as has been shown by Cabana et al. [11
]. However, non-adherence may also be caused by a deliberate decision to counteract the guideline with which the physician may not agree, in general or for a particular patient.
Even though the GP appears to be the main cause of non-adherence to the guidelines, our study clearly shows that other reasons are involved in at least one third of all quality indicators.
Of the non-GP-related causes, adverse drug events and non-compliance appear to be the most important. In chronic care, GPs are confronted with the problem that they have to keep the patient compliant over a long period of time, and that any drug treatment has to match up to other health goals and is influenced by psychosocial problems [16
]. Chapman and co-authors found a sharp decline in drug-adherence to lipid and blood pressure lowering drugs to only 36% within one year, with the greatest drop occurring in the first three months [17
]. The factors determining compliance are manifold: Health education appears to play a crucial role, but other patient specific factors like race, ethnicity, or education are also important [18
]. Sometimes patients do not seem to be aware of their illness, or they accept their chronic disease symptoms as normal, e.g. as a result of ageing. Thus, more than half of the patients with heart failure reported their health to be good even though nearly half of them could not walk a quarter of a mile [5
]. Another problem may be that many drug effects in cardiovascular prevention are hardly noticeable to the patient so that the importance of the medication remains unrecognized. Patients’ fears of adverse drug reactions certainly also play a role in non-adherence to recommended treatments. Moreover, insufficient communication including incomplete patient’s history taking, conflicting information, neglected disagreements, or a disturbed relationship between the patient and the physician may cause non-adherence or non-compliance [19
The fact that about two thirds of the patients did not know why they do not receive a recommended drug points out a significant information deficit. While we would not expect all patients to wish to be informed about treatment options, there appears to exist sufficient evidence that most patients would prefer to be involved in evidence based treatment decisions [20
]. From our study we cannot distinguish whether the information deficit is due to a lack of communication between physician and patient, or to the patient not wanting to be involved. Looking at one of the leading models for shared decision making it takes both the physician’s willingness to share information as well as the patient’s desire to be informed [21
], and we conclude from our study that it seems unlikely that about two thirds of the patients do not want to know about guideline-adherent, evidence-based treatment choices.
About 20% of the patients in our study stated that they do not need a drug they should in fact receive. This might reflect certain knowledge deficits regarding present diseases or risk factors.
Although our study reveals important insights regarding the causes of non-adherence to guideline recommendations, some limitations have to be considered. A major weakness of our study is that the sample size is fairly small. This especially limits the explanatory power of the detailed analysis of single diseases, and even the power of the combined data presented in Figure cannot be considered sufficient due to large confidence intervals reaching zero on the left side.
Even though we chose a random sample of physicians, our study may be biased by a response rate of only 50% on the physician level. On the other hand, this response rate is quite usual in studies involving primary care physicians. The systematic review of Cabana et al. reports a median response rate of 54.5% for studies investigating physicians’ lack of awareness as one of the barriers to guideline adherence [11
It may be assumed that rather motivated physicians who already provide higher quality service to their patients are more likely to participate in a study of quality analysis. But as our main goal in this study was not to quantitatively analyse non-adherence, but rather differentiate the possible reasons for non-adherence, we believe our data to be quite representative.
We visited each general practitioner only one day and included only patients that consulted their doctor at that time. This may bias our results as well, as one may assume that less compliant patients tend to visit their physician only at rare intervals. We tried to overcome this problem by also including patients coming to the surgery only to pick up a prescription.
We used a narrow set of only process quality indicators to judge guideline adherence in drug therapy. This may also be considered a weakness of our study. In diabetes mellitus type 2, for instance, it has been demonstrated that there is insufficient evidence for many widely used quality indicators regarding their predictive power for clinically relevant outcomes [22
]. This appears to be a general problem of using quality indicators, and is not specific to our study.
Last not least our structured interview technique could only obtain categorical data on possible reasons for non-adherence to evidence based guidelines. An in-depth qualitative analysis of individual patients’ reasons for not taking a drug or physicians’ reasons for not prescribing it would be highly desirable and warrants further research.
The strengths of our study are that we examined a representative sample of consecutive patients from both rural and urban areas, and we included all major diseases affecting the cardiovascular system. So far very little is known about the causes of non-adherence to guideline recommended therapy on the patient’s side, and thus this study provides data inspiring further research to improve guideline-adherence in chronic care.