• Safety culture
• Patient safety
• Quality measurement

Primary care in the US model is fraught with safety hazards. Visits are brief and infrequent, patients are largely self-managing, often with multiple comorbid conditions, the extent of healthcare teams varies widely, and the fragmentation of health systems and lack of interoperability among electronic health records (EHRs) means that primary care providers may not have timely, accurate data about patients.1 Despite these multiple vulnerabilities, significant gaps remain in our understanding of the safety of primary care.2–4

The patient safety movement began in acute-care settings, where adverse events resulting from medical care are more immediately apparent. Even though primary care is the cornerstone of healthcare delivery, relatively less is known about the epidemiology of adverse events in primary care settings.5 In an effort to address this gap, Panesar et al6 conducted a systematic review of patient safety incidents in primary care.

This paper makes a major contribution to the field by providing an overview of the burden of patient safety incidents in primary care. They found that patient safety incidents are slightly less common in primary care, around 2%–3% of visits, compared with approximately 10% of hospitalisations.7–9 Given how much larger and healthier the populations using primary care are compared with hospitalised patients, the frequency of safety incidents in primary care is staggeringly high. Fortunately, only a small proportion of these incidents result in severe harm.

This systematic review highlights multiple challenges in studying the safety of primary care. First, even more than a decade into the patient safety movement, definitional challenges remained. Panesar et al6 defined patient safety incident as ‘any unintended or unexpected incident that could have or were judged to have led to patient harm’. Within this broad definition, however, the authors chose to include only incidents of commission rather than omission. In effect, they counted events where the wrong thing was done, but did not count events where the right thing was not done. The fast pace and frequent interruptions associated with primary care are known to lead to errors of omission,10 and omissions are a major culprit in missed and delayed diagnoses,11 which Panesar et al found to be among the most harmful of primary care safety incidents. Therefore, this analysis likely shows us only the tip of the iceberg. The included studies under-represent the frequency of patient safety incidents, and may especially undercount diagnostic errors. It is critical that, going forward, safety surveillance efforts in primary care include incidents involving errors of both commission and omission.

The ascertainment methods for patient safety incidents require further examination.12 Most of these studies employed record review, while other used incident reporting systems or surveys. Each of these ascertainment methods has limitations. Record review leads to lower estimates of incidents because of suboptimal documentation13; incident reporting systems are underused,14 particularly by physicians15 and surveys can include incidents which are not related to safety per se.16 Future studies should employ multiple ascertainment methods for primary care patient safety incidents and contrast the resulting estimates.

Evaluating the extent of harm to primary care patients remains a thorny issue. Expert record review does not always yield agreement about harm.17 While the authors used a clear definition from the UK National Patient Safety Agency,18 the variation in the underlying studies suggests that the definition may not have been applied consistently. One example of an incident without harm is an illegible handwritten prescription which does not impact a patient—perhaps because the pharmacist contacted the physician to clarify. However, an erroneous/harmful medication prescribed by a physician but identified and intercepted by a pharmacist, therefore, not reaching the patient, also represents an incident without harm. These two incidents are conceptually distinct. Developing effective interventions requires precision in describing harm.

Many things can go wrong in primary care. The types of incidents that emerged from this synthesis are medication events, diagnostic errors and communication failures, which I infer includes handoffs among outpatient providers. These results underscore the importance of these three areas as the three pillars of primary care safety. All require substantive investment in improvement and evaluation. While electronic prescribing has eliminated certain medication-related incidents,19 abundant evidence indicates that medication use remains a locus of safety problems in primary care.20 Regarding diagnosis, a recent Institute of Medicine report estimated that every American will experience a missed or delayed diagnosis.21 ,22 The frequency of lasting harm in delayed or missed diagnoses adds to the concern about addressing this often-overlooked safety problem.

Communication failures likely encompass two critical areas for the safety of primary care—handoffs between outpatient providers, who rarely share EHR systems or processes,23 and communication with patients. Because patients with chronic conditions are often expected to perform complex self-management tasks independently,24 suboptimal communication with patients puts them at risk of harm.25 These incidents demonstrate the need to embed effective, health literacy and language-appropriate communication in routine primary care.26–28

Panesar et al6 synthesis of patient safety incident rates illuminates the agenda for patient safety in primary care. The field urgently needs consensus around definitions of events, harm and preventability that can be implemented with consistency across multiple settings. These definitions should extend from research to practice. We need multicentre, prospective, epidemiological studies using multiple surveillance methods in order to clearly understand the incidence and prevalence of safety incidents in primary care, including both omission and commission events. Improving communication requires a heightened focus on patient engagement and communication needs and preferences. Interventions that seek to reduce harm to patients should focus on both prevention of adverse drug events and improving diagnosis. This research agenda cannot proceed without substantial investment in primary care patient safety research and quality improvement efforts. This investment should come from health systems, payers and research funders in order to improve safety for all populations receiving primary care.

Because most primary care safety incidents do not lead to immediate harm, they do not seem to carry the same urgency as adverse events in the hospital. But, even if the consequences of a missed cancer diagnosis or dangerous medication combination error take months to years to manifest, they remain just as morbid and just as important to patients as any preventable adverse event in the hospital. Even if most patient safety incidents in primary care carry a low potential for lasting harm, as Panesar et al show, the sheer volume of service delivery translates into a substantial public health burden from patient safety incidents in primary care. The time has thus come to broaden our focus in patient safety accordingly. The tip of the iceberg is a place to start.