Reasons for and consequences of missed appointments in general practice in the UK: questionnaire survey and prospective review of medical records

The study was conducted over a three week period in 2001 in seven practices in West Yorkshire. These differed in terms of size, location, annual consulting rates, organisation of clinical workload, and missed appointment rate. Whilst all were members of a research network, it has recently been demonstrated that research practices do not differ from others in terms of their demography or morbidity [13].

All adult patients who missed an appointment and patients in the comparison group (the next adult seen in the same surgery session to control for: doctor-related factors, climate factors and practice factors) were sent a pack within 24 hours of the appointment, comprising a short and neutral invitation letter, and a questionnaire with open and closed questions about the missed appointment (cases only) and appointments in general. Patients with severe mental impairment, learning difficulties, severe emotional distress, or terminal illness were excluded. Two reminders were sent to initial non-responders.

A second mailing was sent to responders comprising a GHQ-12 [14], and a consent form to review medical records. Multi-centre and local ethical approval was received.

The first part of the questionnaire aimed to empower and engage the patients by asking their views on how practices could help patients keep appointments (data not reported in this paper). The second part (cases only) asked directly the reason for making the appointment, and the reasons for missing it. The third section (cases only) explored the reasons behind the missed appointment by asking their agreement with sixteen statements.

Returned GHQ-12s were scored according to the manual [14]. Patients scoring three or more were sent a letter explaining that they may be suffering from mental health problems and were advised to seek help from their GP. For those providing written consent, data were abstracted from medical records three months after the index appointment, relating to whether they consulted for the stated reason of the missed appointment, the number of days from the missed appointment to the next consultation, and the number of missed and kept appointments in the twelve months prior to the index appointment.

The sample size calculation was determined by the difference in GHQ-12 scores between cases and the comparison group. In order to identify a high and low GHQ score difference of 20% between the groups at 5% significance level and 80% power, 200 pairs of patients and controls would be required. To allow for a likely response rate of approximately 60%, we estimated a sample size of 360 pairs.

Free text data relating to reasons for missing appointments were read independently by three of the researchers (MH-G, OD and RDN) and the main themes discussed. It was apparent that for each patient, there was one main reason. A categorisation emerged that was applied across the data. This was then refined further and the final categories agreed. Comparisons between patients who missed appointments and their controls were assessed using conditional multiple logistic regression for matched pairs. Data relating to the numbers of missed and kept appointments for patients who were not registered for the entire year prior to their missed appointment were adjusted to allow for the proportion of the year that they were registered and rounded to the nearest integer.

Of the 386 patients who missed appointments 122 (31.6%) returned a valid questionnaire. Table 1 shows the responses to statements provided in the questionnaire concerning factors contributing to the missed appointment. Over 40% said that they forgot the appointment and a quarter that they tried very hard to cancel the appointment or that it was at an inconvenient time. A fifth reported family commitments or being too ill to attend. Table 2 shows the categorisation of free text reasons for each of the 122 respondents. The largest categories were 'misunderstandings and mistakes', 'illness or personal circumstances', 'forgetting', and 'other commitments'. Of the misunderstandings and mistakes, the largest sub-category was 'by the practice', and included 'being unable to get through', 'had cancelled', 'did not have an appointment', and 'told wrong time or date'.

Of the 122 patients who missed an appointment and responded, 57 (47%) consented to medical records review. Age and gender distributions between those who gave consent and those who did not were similar (both p values > 0.2). Of these, 52 (91.2%) patients subsequently consulted within three months of the index consultation. One third were in the first week and over half were within three weeks. 33 (63.5%, 95% CI 49.0% to 76.4%) consulted for the stated problem that was going to present in the missed consultation. Nineteen patients (36.5%, 95% CI 23.6% to 51.0%) had no record of consulting for the same reason, and it was unclear for five patients.

The overall response rate among the comparison group (223, 58%) was greater than for cases, and there was a tendency for a greater proportion of the comparison group who responded, to also respond to the GHQ-12 (65.4% versus 50.8%), and to consent for medical record review (61.0% versus 45.1%). There were 27 case-control pairs who responded and provided data on the GHQ-12 and 23 pairs who responded and provided consent to medical records.

The odds of missing an appointment was lower for women compared to men (odds ratio (95% confidence interval) 0.67 (0.19, 2.36)), and the odds of missing an appointment decreased with increasing age (odds ratio (95% confidence interval) for an increase of 1 year in age 0.95 (0.91, 0.99)) though due to small numbers these differences were imprecise and the gender difference did not reach statistical significance at the conventional 5% level. The odds of missing an appointment were greater among those who had missed at least one appointment in the previous 12 months (5.88 (0.83, 41.36)), though again due to low response, this did not reach statistical significance at the conventional 5% level. This association was not substantively changed with adjustment for age, sex and the number of appointments made in the previous 12 months. There was a slight tendency for the odds of missing an appointment to be greater among those with a high GHQ-12 score (≥ 3), though because of small numbers the estimates were imprecise. With adjustment for age and gender the odds ratio (95% confidence interval) for missing an appointment comparing those with a high to a low GHQ-12 score was 1.14 (0.41, 3.15).

The age distribution of the comparison group was in keeping with national figures for those attending general practice, indicating that the practices participating in this study are representative of general practice as a whole [15]. One of the main strengths of this study is the investigation of an area which is important to primary care but which has been rarely investigated, perhaps because of anticipated difficulties. Further, we have attempted to improve upon previous studies with a prospective design, comparisons with a comparison group and use of a neutral place for correspondence. The main weakness of the study is the low response and the differential response between patients and the comparison group. A selection bias may exist between those who responded with respect to the outcomes of interest; this may be because of both social desirability bias and post-hoc rationalisations in their written responses. Responders may be over-represented by those whose missed appointment was a genuine practice mistake, and those who were less obliged to operate within the practices' booking systems being less likely to respond. Assuming that all 68% of non-responders did not miss their appointment because of forgetfulness or practice misunderstandings, then interventions aimed at dealing with these problems would only be able to reduce the overall missed appointment rate by a small amount.

Within the category 'misunderstandings and mistakes', the largest sub-category was 'by the practice'. Further, in the response to the pre-defined categories, 30% stated that they had tried to cancel their appointment. A small number of patients reported that they had either cancelled their appointment or indeed kept their appointment. These findings were not verified by the medical records, therefore it is difficult to know whether these represent practice mistakes of post-hoc rationalisations of behaviour. Whilst this may reflect post-hoc rationalisations or may be exaggerated by selection bias, there is consistency between the two questions and the results suggest that improvements in practice communication systems could reduce missed appointments. Forgetfulness was another important cause. One practice strategy that may reduce this problem would be the use of aide-memoirs. However, intervention studies are required to demonstrate their effect. In a parallel study we have assessed the perceptions of primary care health professionals of the reasons for missing appointments [9]. Interestingly, practice staff tended to blame patients for missed appointments and dismissed the idea that practice factors may contribute. The results of this study suggest that practice factors contribute to some missed appointments suggesting that considering practice level interventions (e.g. making it easier to cancel appointments, and greater convenience of appointment times) may be useful.

The finding that virtually all patients who missed an appointment did subsequently consult within a three month period has not been previously demonstrated. In fact, over half of those who responded consulted within the following two weeks of the missed appointment. Again selection bias may have influenced this result if those who did not respond were different to those who did and were less likely to subsequently attend. This study was unable to determine whether those with psychiatric morbidity are more likely to miss appointments; and whilst further research in this area is required to determine this, a recent paper has found that psychological morbidity did affect the likelihood of missing appointments [7].

There are several implications of these results for future research. This work adds to the literature confirming that more work needs to be done to engage people who miss appointments with research in a more meaningful way. Creative approaches, in terms of accessing and engaging patients who have missed appointments in research, and of using appropriate methodologies to answer important questions need developing; qualitative methods may be best at exploring the phenomenon from user perspectives and would help to unravel the complex relationships surrounding consultation etiquette and patients' and practitioners' behaviour. Closer involvement of users might inform this process [16], as may the use of inducements. One option would be to recruit patients when they next consult in primary care, as we have shown the vast majority do consult again soon. If research was then conducted in a neutral environment, the use of practice members to recruit participants need not bias the responses. The lessons that we have learned have implications for future research for others who are developing work with other 'disengaged' or potentially vulnerable groups of patients. There is still a need for health professionals to maintain their concern for the health of patients who miss appointments, since this study does not exclude the possibility that they may have unmet mental health problems.