• clinical governance
• general practice

Several arguments can be readily raised against monitoring outcomes in primary care. Many patients attending primary care have minor self-limiting illnesses, and the only outcomes that might be of interest for monitoring would be significant adverse events such as drug reactions. At the practice or practitioner level the numbers of patients with significant disorders are small, and distinguishing the impact of variations in quality of care from case mix and random variation is difficult. Research evidence is often incomplete and the relationship between the process of care and its outcome is not well established. The interval between the delivery of healthcare interventions and eventual outcome can be many years, as in the case of illness prevention strategies. Primary care is provided by teams, and it is frequently impossible to ascribe a particular outcome to the care provided by an individual doctor or nurse. Some family doctors would also argue that their role is more complex than simply the achievement of desirable clinical outcomes following illness episodes; family doctors have long term relationships with their patients, and ultimately all their patients die.

Intermediate outcomes have offered one approach for responding to these problems. Monitoring of immunisation rates has long been routine in the healthcare systems of many countries, and in recent years disease control markers such as glycated haemoglobin or blood pressure have begun to be included in monitoring schemes. The validity of intermediate outcome measures depends in large measure on the completeness of the data. While information about immunisation in a population of children may be relatively easy to compile, the creation and maintenance of an accurate list of all patients with diabetes or ischaemic heart disease is much more difficult. When financial incentives are attached to the achievement of intermediate outcome targets, the problem of obtaining complete and accurate data can increase. Monitoring of deaths avoids some of the problems inherent to monitoring of intermediate outcomes. In most countries, central systems to collect and record information about all deaths are in place, and the data are much more likely to be complete. However, the disadvantages of monitoring mortality in primary care have been regarded as insurmountable, and include the small annual number of deaths in the small populations of primary care teams and practitioners, the delay between the delivery of care and death, the sharing of care between different practitioners and between primary and secondary care, and the impact of case mix. Beaumont and Hurwitz, in this issue of QSHC, have identified another problem.¹ General practices have considerable difficulty in obtaining details about all the deaths among their patients, and they recommend that a central system should be established to fulfil this task. I have not been able to identify any developed country in which death rates in populations cared for by primary care teams are routinely studied to monitor performance at the level of the team.

“ . . . always monitor key outcomes”

The discovery that a general practitioner was able to murder more than 200 patients over a period of 20 years and not be detected until he decided to forge the will of one of the patients he had killed has caused practitioners in the UK to question the belief that monitoring mortality in the small populations of primary care is not worthwhile.² Family doctors in other countries would be wise to take note of the UK experience and consider whether they too should rethink the widely held view on the use of mortality data. A monitoring system is to be set up in the UK following a recommendation made as a consequence of a review of the clinical practice of the GP murderer.³ Problems will be encountered, but new methods of analysis suitable to small numbers will overcome some of these. Local knowledge of the populations of patients served by different primary care teams and the patterns of work of different doctors will help to overcome other problems. If information about other outcomes such as stroke or myocardial infarction were included in the monitoring system, it may prove possible to monitor the impact of clinical policies and identify primary care teams that could do more to prevent death and other serious non-fatal events. The prospect of managing performance in primary care to improve key outcomes is attractive but speculative; more evidence about the sensitivity of monitoring and its feasibility in routine use is required. Nevertheless, there is an important lesson for all with an interest in safety and quality improvement—always monitor key outcomes. If you don’t, you won’t know when the outcomes are poor. Why then should your patients trust you?