This study provides an insight into patient safety incidents through medical record review in general practices. A total of 211 patient safety incidents were found to have occurred in 8,401 contacts with the GP practice (in 1,000 patient years). Of these 211 patient safety incidents, 58 affected the patients and seven of these were associated with an unplanned hospital admission.
Other studies of the occurrence of adverse healthcare events reported widely varying prevalence rates. These studies mostly involved incident reporting, although patient reported incidents or malpractice claims have been researched as well. None of these studies undertook a medical record review. Moreover, in our study we only included preventable patient safety incidents, while other studies also included non-preventable incidents. These are important differences, which are likely to yield different numbers and types of incidents. There are also differences between primary care and other sectors, which complicates comparison. In the United States, 33 primary care practices (475 clinicians) reported 608 incidents over a two-year period . Another study showed 100 incident reports by healthcare workers in a one year period (with 25,000 visits) in an ambulatory care setting . A literature review of studies on medical errors in primary care showed a prevalence of 5 to 80 times per 100,000 consultations . The present study showed a much higher rate, namely 2,512 patient safety incidents per 100,000 consultations (95% CI: 2,198 to 2,869). The present findings could reflect the use of a broad definition of the term 'patient safety incident'. In the present study, most (72.5%) of the patient safety incidents indeed had no tangible impact on the health of the patient. If we only consider those patient safety incidents with tangible consequences for the patient, we find a prevalence of 690 patient safety incidents per 100,000 consultations (95% CI: 534 to 891) (0.69% of the patient contacts or in 18.6% per patient per annum), which is still considerably higher than reported in other studies. The large gap between the present data and the numbers published by Sandars in 2003 can be explained in several ways. Sandars' review of the literature mostly included studies that were based upon the reporting of health professionals. While all methods for the measurement of patient safety may involve potential bias [8, 22], one could conclude that the direct review of a random sample of medical records could be the most thorough method for the measurement of patient safety incidents. Back in 2003, Sandars also already advised: 'to maximise reliability of error reporting, it is beneficial to obtain data from a second reporter rather than relying on the physician alone.'
The health consequences of the present findings at a national level are potentially quite large. For example, our findings suggest that about 60,000 hospital admissions per year are potentially related or at least partly related to patient safety incidents in primary care (95% CI 25,776 to 140,325). There were 1.8 million hospital admissions in the Netherlands in 2007. This estimate lies within the range of previous studies concerned only with medication errors in the Netherlands and showed 41,000 Dutch hospital admissions per year to be related to medication errors, with 19,000 or almost 50% of these 'severe' medication errors potentially avoidable .
From the perspective of the individual patient, however, general practice appears to be safe. Research in hospitals shows one or more patient safety incidents to have occurred in 5.7% of hospital stays, with a preventable adverse event occurring in 2.3% of hospital stays. Other hospital-based studies tend to have even higher incidence rates of approximately 10% . Nevertheless, the occurrence of 1,482 to 2,032 potentially preventable deaths in Dutch hospitals per year is the result of these patient safety incidents in hospitals [12, 24]. In contrast, in the present study, no adverse events were found to lead to a preventable death. Although corresponding percentages of patient safety incidents were found in the GP and hospital settings, the potential consequences of the patient safety incidents in general practice were much less serious than those of the patient safety incidents in hospital. This probably reflects the generally lower risk of the majority of interventions conducted in general practice, the fewer number of transfers of patients between health professionals in general practice, and the generally healthier status of patients in the GP setting, as opposed to the hospital care setting.
The results of the present study are of particular relevance to countries with a strong primary care system. About 95% of the health problems of patients in the Netherlands are fully managed by GPs in primary care. The threshold for hospital admission is probably higher compared to countries with less well-developed primary care systems. This could constitute a potential safety risk, as the family practitioner must make clinical decisions with the aid of only a few diagnostic possibilities (e.g., no x-rays, frequently no EKG possibilities). Conversely, this same threshold could actually reduce the risk of iatrogenic damage; fewer false positive test results could occur as a result of less testing in the primary setting and less 'over-testing' of the patient could occur in the primary care setting, compared to the hospital setting. The most serious patient safety incidents in our study were found to be related to clinical decisions in which a 'wait and see' approach was inappropriately adopted. For example, when no further additional testing was conducted for a patient with chest pain. This finding is also in line with the results of other studies that underscore the significance of diagnostic errors .
An exploratory analysis of the patient safety incidents showed those patients who visited the primary care practice more than 11 times a year to have a heightened probability of experiencing a preventable adverse event. In a multivariate model, moreover, other variables such as age, gender, polypharmacy, and patient-at-risk lost their significance when included with frequency of practice consult. In other words, the most common health risk factors were not related to the number of patient safety incidents, while frequency of primary care practice visit was. We suggest that the chances of a preventable adverse event are the same for every practice visit, but increased practice visit additively increases the probability of a preventable adverse event due to so-called chance capitalization. One study shows patients with a high frequency of practice visits to be mostly female, have a BMI >30, have alcohol abstinence, and low patient satisfaction, for example . Of course, another -- still unknown -- variable might account for the association.
In our opinion, further research should focus on two points. First, the diagnostic process and the wait and see approach, which is an important tool in general practice, and second, education on patient safety and improvement on this subject.
In sum, serious patient safety incidents appear to have lower prevalence in the general practice than in the hospital setting. Also, the outcomes of patient safety incidents, when they occur, appear to be less serious in the general practice than in the hospital setting. The general practice setting thus appears to be a relatively safe place for the patient, but awareness of harm should nevertheless be enhanced given the potentially detrimental consequences of such harm when it does occur.
Each of the methods available to determine the prevalence of patient safety incidents has its difficulties. The literature shows little overlap in the different methods used to document the prevalence of patient safety incidents . Retrospective studies of patient records currently offer the best means to assess the prevalence of patient safety incidents . Nonetheless, the reporting of patient safety incidents by healthcare professionals may be more appropriate for attaining a more in-depth understanding of patient safety incidents. Even so, many of the reported patient safety incidents stem from organisational and communication problems. There is also a suspicion of underreporting medical errors by healthcare professionals . The generalisability of the present findings could also be limited by the relatively low number of health professionals and primary care practices involved in the study.
The reliability of reviewing patient records could be problematic. In our study however, the inter-rater agreement (κ values) was found to be reasonably good. It thus appears that our level of agreement was comparable, or better than the level of agreement found for similar empirical research conducted in a hospital setting [12, 16]. The retrospective interpretation of patient records could nevertheless be biased by hindsight .
Finally, in the root cause analyses, we noticed that mostly human and organisational factors played a role in the occurrence of patient safety incidents in primary care. It is known that the underlying causes of patient safety incidents could also be largely technical and system-related . Patient records generally provide insufficient information for a thorough root cause analysis. The present study would therefore have been strengthened if in-depth interviews with family practitioners had been conducted to explore the roles of various contributory factors. This was unfortunately not feasible, due to time and financial constraints.