The disappearance of the “revolving door” patient in Scottish general practice: successful policies

Background We describe the health of "revolving door" patients in general practice in Scotland, estimate changes in their number over the timescale of the study, and explore reasons for changes, particularly related to NHS and government policy. Methods A mixed methods predominantly qualitative study, using a grounded theory approach, set in Scottish general practice. Semi-structured interviews were conducted with professional key informants, 6 Practitioner Services staff who administer the GP registration system and 6 GPs with managerial or clinical experience of working with “revolving door” patients. Descriptive statistical analysis and qualitative analysis of patient removal episodes linked with routine hospital admissions, outpatient appointments, drug misuse treatment episodes and deaths were carried out with cohorts of “revolving door” patients identified from 1999 to 2005 in Scotland. Results A “revolving door” patient is removed 4 or more times from GP lists in 7 years. Patients had complex health issues including substance misuse, psychiatric and physical health problems and were at high risk of dying. There was a dramatic reduction in the number of “revolving door” patients during the course of the study. Conclusions “Revolving door” patients in general practice had significant health problems. Their numbers have reduced dramatically since 2004 and this probably resulted from improved drug treatment services, pressure from professional bodies to reduce patient removals and the positive ethical regulatory and financial climate of the 2004 GMS GP contract. This is a positive development for the NHS.


Background
Interest in "revolving door" patients, that is those who are repeatedly removed from successive general practice (GP) lists, arose from pilot work on how patients achieve registration with practices in the city of Glasgow. General practices in the United Kingdom operate a geographical list system which defines their patient population. Being registered with a general practitioner (on a list) is necessary to access most National Health Service (NHS) facilities. We wished to find out more about this group of patients who appeared to be systematically excluded from a health system that is internationally lauded as providing trust, coordination, continuity, flexibility and coverage, irrespective of health status or ability to pay [1].
General practices in the UK have a right to remove patients from their list for a variety of reasons including "break down in the doctor-patient relationship" or "violence". "Moved out of the practice area," is more commonly used but this category is rarely relevant when considering "revolving door" patients. There is a literature on patients who have been removed once from GP lists but it has either excluded repeatedly-removed patients from analysis [2] or has recommended management strategies without any consideration of their characteristics or the reasons for their repeated removal [3].
In this paper we develop a definition of "revolving door" patients in general practice and describe their health. We also report on changes in the number of "revolving door" patients in Scotland over the timescale of the study and the possible NHS and government policy reasons for this. The explanations as to why patients may become "revolving door" patients and the implications for our understanding of doctor-patients relationships will be considered in a future paper.

Methods
In 2006, six semi-structured interviews were carried out with Practitioner Services staff who administered the GP registration system across Scotland and with two GPs whose managerial or clinical role meant they had worked with "revolving door" patients. Data from these interviews guided the development of a definition of a "revolving door" patient and this definition was applied to the patient removal data (for reasons other than change of address) obtained from the Community Health Index (CHI) from 1999 to 2005. Secondary care data on hospital admissions, outpatient attendances and drug misuse treatment episodes experienced over the life course were linked by NHS Information Services Division (ISD) Scotland to a sample. The results for the adults are presented in this paper. Additional file 1: Appendix 1 describes the statistical analysis in detail.
Drawing on cognitive psychology script theory about clinical decision making [4], the sensitizing concept (idea that helps to shape theory generation) [5] of "patient scripts" was used to categorize patients' predominant health issues into categories that described the cohort overall and also guided the direction of statistical analysis. Additional file 1: Appendix 2 describes the use of "patient scripts" to do this.
A further four semi-structured interviews were conducted in 2010 with GPs in health board areas where "revolving door" patient prevalence was high. Summaries of patient removal data from ISD Scotland (1999 to 2011) were also reviewed and analyzed.
Our predominantly qualitative analysis using mixed methods was informed by the grounded theory approach attributable to Charmaz [5]. The results were integrated in a dialectic way; that is they were compared to seek and explain differences between them [6].

Results
Definition "Revolving door" patients in general practice were described by the professional key informants in 2006 as a small group of patients that professionals working in primary care would recognise. In 2010 one GP described coming across 20-30 "revolving door" patients during a 15 year partnership in an urban area that used to generate a lot of repeat removals (GP respondent 3, (GP3)).
In 2006 Practitioner Services respondents agreed that a patient who had been removed once or twice was not a "revolving door" patient, but three removal episodes was "starting to look like a problem". There was a range of views on removal frequency. Respondents made a distinction between two groups of "revolving door" patients; "fast revolvers" who were regularly and routinely removed as frequently as every seven days and "slow revolvers" who were repeatedly removed, but after months rather than days or weeks.
We developed several versions of the definition in an attempt to describe these qualitatively derived categories so included all patients removed four or more times during the study interval. We tested categories based on the median number of days patients spent on a GP list. Those with a median of 0-100 days on list were "fast revolving door" patients, 101-180 were "medium revolving door" patients and those with medians of more than 181 days we called "slow revolving door" patients. Table 1 shows there were no substantial differences in demographic characteristics (sex, age, marital status and deprivation) between these categories, although the proportion of males was slightly lower among patients who had stayed on a GP list with a median of more than 180 days compared with faster revolving patients (59% v 69%; P = 0.045). Additional file 2: Appendix 3 summarises additional tests that we carried out across the  data to compare the "fast", "medium" and "slow" revolving door categories. We could find no additional statistical differences between the groups from these tests either.
Because there was no difference between these groups our final working definition of a "revolving door" patient was a patient who was removed four or more times from a GP practice list in seven years.
There were demographic descriptive data for 555 adult "revolving door" patients, and health service linked data for 410 patients. 351 patients were included in the qualitative analysis. Additional file 2: Appendix 4 summarises the three samples of "revolving door" patients. They were considered representative of the "revolving door" patient cohort as a whole.
Most (83%) "revolving door" patients were removed between four and seven times (range 4-92), and most (67%) were male. The median age at first removal was 31 years (range 17-88). Thirty-five percent of "revolving door" patients lived in the most deprived decile and 87% in the more deprived half of Scotland (decile 6-10) [7]. Eighty-four percent were not married at their first removal (compared with 51% of the general Scottish population aged 30-34) [8].

Morbidity
Eighty-six percent of the 410 record-linked "revolving door" patients had at least one hospital admission (median 9 admissions, range 0 to 295) before 2011. The reasons for admission are summarized in Table 2.
Fifty-one percent had taken an irregular discharge during a hospital admission (median 1, range 0 to 45). Ninety-nine percent had had a hospital outpatient appointment (median 15, range 0 to 249) and 92% of patients had missed outpatient appointments (median 7, range 0 to 146).
There were no substantial correlations between numbers of removal episodes and other patterns of health service activity, including irregular discharges (Spearman's ρ = 0.02, P = 0.743) and number of admissions (Spearman's ρ = 0.07, P = 0.170). There was however a weak positive correlation with the number of outpatients appointments (Spearman's ρ = 0.11, P = 0.032), although not with the number of missed outpatients appointments (Spearman's ρ = 0.07, P = 0.141). There was a tendency for hospital admission dates to be close to removal dates (odds = 1.25, P = 0.008) but not to reregistration dates (odds = 1.16, P = 0.084) than could be expected by chance (a health service date was defined as close to a practice transfer date if fell in the first or last 25% of the period between two practice transfer dates; see statistical methods). Dates of drug treatment episodes were close to both removal (odds = 1.26, P = 0.049) and registration dates (odds = 2.05, P < 0.001). Outpatient appointment dates were not associated with removal (odds = 0.98, P = 0.740) or registration dates (odds = 1.03, P = 0.701).
From all the collated data sources it was determined that 84% of patients had been dependent on substances. The majority had evidence of dependency on opiates but patients with alcohol dependency were also numerous. Alcohol-related admissions were important for many patients in the cohort.
Forty-eight percent of the cohort had evidence of self harm and 18% had a definite personality disorder diagnosis. There was a low prevalence of 'severe and enduring' (chronic psychotic or severe mood disorders) mental health problems.
Qualitative analysis of the "revolving door" patients gives an overview of the main health problems of the sample and is summarized in Table 3 together with examples of patient profiles in each category. Patient profiles which were coded "no clinical code possible" had insufficient evidence from the linked secondary care and drug misuse data to be able to determine their main health problems. We did not have access to primary care records. Figure 1 compares mortality rates from 1999 to 2008 between the 410 record-linked "revolving door" patients and the general Scottish population. It shows the standardised mortality ratio (SMR) in the "revolving door" cohort, overall and within age, sex and deprivation subgroups, relative to the general population in Scotland in 2004 [9]. The SMR is defined as the number of deaths observed among "revolving door" patients for every 100 deaths in the general Scottish population. The SMR estimates were adjusted to take account of differences in mortality rate between the two populations due to differences in the distributions of age, sex and deprivation.

Mortality
The overall SMR for the "revolving door" patient cohort was 333 (95% CI 264 to 415) for every 100 deaths in the general population. In other words, there were around three to four deaths in the "revolving door" cohort for every death in the general population. Among younger "revolving door" patients (<45 years), the SMR is 558 (95% CI 401 to 726), more than double the SMR of patients aged over 45 (relative risk 2.5, 95% CI 1.6 to 3.7), so the death rate is approximately six-fold higher than in the general population. There were no significant differences in SMR between males and females or the most deprived 20% and the least deprived 80%.

Apparent disappearance
That "revolving door" patients were reducing in number was evident when the first interviews were conducted in 2006 as described by this key informant: I (interviewer): ". . .this idea of "revolving door" patients do you think that's a valid one?" R (respondent): "I might have a couple of years back but I don't think so much now. The GP contract changed in 2004 and my allocations have literally gone down to zilch so the contract has been great for me. I do have the offenders, my ones that are continually going round the system but in saying that they stay longer with a practice now before they are put off; they are no longer Figure 1 Estimates of the standardised mortality ratio (SMR; the number of deaths observed per 100 expected) among the 410 record-linked "revolving door" patients relative to the general Scottish population in 2004, overall and in subgroups of age, sex and deprivation. Error bars represent 95% confidence intervals. The SMR estimates were adjusted for differences between the "revolving door" cohort and general Scottish population in the distributions of age (by 10-year age bands), sex and deprivation (by SIMD decile). The dashed line at SMR = 100 represents equivalence in mortality rate between the two populations; that is, if the adjusted death rates were the same in the two populations, the confidence intervals would be expected to overlap the dashed line.

Physical illness 4%
Female patient in 20s, 5 admissions. She had nausea and vomiting, biochemical and coagulation problems. The underlying diagnosis was unclear. Missed 5/6 outpatient appointments including psychiatry appointments. Removed 5 times from GP lists.

No clinical code possible 29%
Male patient in his 30s, 10 admissions. Open wound to forearm and no other recorded information on other admissions. 5 outpatient appointments in oral surgery, orthopaedics and ENT with 1/5 missed appointments. Removed 5 times from GP lists. a seven day removal; so it's working for me." (Practitioner Services respondent (PS4)) The scale of the reduction in "revolving door" patient numbers became apparent in 2009, (four years after the previously described "revolving door" patient cohort) when no patients meeting the definition of a "revolving door" patient (as defined earlier) were eligible for recruitment into a patient experience study. Figure 2 plots all removal episodes for reasons of breakdown in doctor patient relationship or violence (excluding geographical removals) by Health Board in Scotland between 1999 and 2011.
Glasgow (and Clyde) demonstrates most removal activity. It is the largest Health Board in Scotland and has the highest proportion of patients living in deprived areas [10]. Table 4 shows the Scottish data for repeat removals from 1999 to 2011. These are calculated within-year so patients who may span a number of years to reach the definition of a "revolving door" patient (including patients with one removal episode per year over four years) are under-represented in these data.
Had it been possible to include the repeat removal data about individual patients for Glasgow and Clyde Health Board from 1999 to 2004 the trend downwards may have been steeper.
The professional key informants described three reasons why they thought the generation of "revolving door" patients had reduced so dramatically over time.
The first was the introduction of the treatment of problem drug use and subsequent development of services. A GP respondent gave a bleak description of the early years of the drug injecting epidemic and then how that changed: "..it really kicked off about 92, 93, a lot of people started appearing, we had no training in it, we didn't know what to do. GPs didn't know what to do, there was no hospital base, there was an alcohol service but there wasn't a drug service and more people were appearing and we didn't know what to do with them. Over time, some of these patients became so insistent and abusive and demanding of practices that eventually they would, we would try our best with them but they would cross a line. . . .when we got a drugs service which was effective and people were getting into treatment, and they were being stabilised, then a lot of these patients' problems disappeared" GP4.
The second reason was the influence external organisations such as the Royal College of General Practitioners and the Health Ombudsmen had, by discouraging GPs from removing patients from their lists.
The third reason was the impact of the 2004 General Medical Services (nGMS) GP contract which changed the way that practices worked and their payment mechanism. This was considered by all key respondents to have had a large positive influence on practice removal activity and the production of "revolving door" patients. The most important aspects were thought to be the non discriminatory tone and accountability for removal decisions that the contract introduced.
A strong theme that emerged from the GP professional key informants was that when "revolving door" patients stopped "revolving" and settled with a practice they remained challenging to work with. There was limited evidence that a very small number of patients was still being removed, but at a much slower rate than previously. This is illustrated by the following quote from a GP interview from 2010. The practice had kept the patient registered for nine months before this removal: R: "There would be one that I could think of that most recently left with mild learning difficulties and significant mental health. . .who again had the difficult way of interacting with the staff, out of hours, and inappropriate requests for things that were insoluble. . .Unfortunately her [relative] verbally, well no physically threatened [an ancillary member of staff], tried to run [them] over; which was something that we couldn't really tolerate. And so because he drove her here, all the time on a daily basis; generally that it was something we could not sustain. So she was already on a warning for behaviour and she apologised for it; her behaviour about verbally abusing several members of the reception staff at the front door as they left to go home from work. . . .she crossed the line it was just unacceptable. . ."GP3.

Discussion
This study defined a "revolving door" patient as one who was removed four or more times from GP lists in seven years (excluding those removed for having moved out of the practice area). "Revolving door" patients in this cohort had substance misuse problems and a mixture of psychiatric health problems which included opiate dependency, harmful alcohol use, self harm and for some a diagnosis of personality disorder. Physical health problems were important too, along with being a victim of violence. There was a high risk of dying in this cohort, with one in six patients having died by the time the study concluded despite this being a predominantly younger adult age population of patients. There was no link between removal rates and use of health services although patients being admitted to hospital tended to be close to removal dates and drug treatment episodes tended to be close to re-registration dates.
There was a dramatic decline in the number of patients who became "revolving door" patients over the time frame of the study and this was considered to be due to changes in the way the NHS worked with patients. Developments in problem drug use treatment came early and more recently the expansion of community treatment services which included integrated working between GPs, community addiction teams, hospitals and prisons, is thought to have led to further improvements in stability of treatment. Hence it is postulated that patients whose primary reason for becoming a "revolving door" patient was their difficult interaction with GPs about their drug misuse treatment, stopped "revolving". There was pressure too from professional bodies to reduce patient removals and the final change appears to be the positive ethical, regulatory, and financial climate of the 2004 nGMS GP contract.
The main strength of this study was that it used mixed methods. This allowed us to contextualise data, direct data collection, and highlight strengths and weaknesses of the data sources. It explored and then represented the complexity of the topic of "revolving door" patients and