Understanding variation in patient care: A qualitative study of hospital (non-ST elevation myocardial infarction) practices

Design

Qualitative design, including observations of NSTEMI care, interviews with clinical staff and patients, as well as scrutiny of medical records.

Hospital sample

We used the Acute Coronary Syndrome (ACS) registry for England and Wales (Myocardial Ischaemia National Audit Project, MINAP, Herret, Smeeth, Walker & Weston, 2010) to identify a sample of 30 hospitals in the top or bottom third of 30-day case-mix adjusted mortality for ACS. We purposively sampled eight hospitals: four from the top tertile and four from the bottom; they varied in coronary interventional facilities, teaching status, volume of cardiac admissions and geography. We piloted the field work methods in two hospitals before commencing on the main sample of eight hospitals. This pilot work gave us a greater insight into the complexity of NSTEMI care and, as a result, led us to focus exclusively on NSTEMI care in the main study. In total there were ten participating hospitals with variation in their organisation of care for NSTEMI patients (see Table 1). As a condition of the research the hospitals were not aware of their selection based on performance status for 30-day mortality. The research team was also blinded to hospital performance status until the analysis of qualitative data was completed.

Consent

Before any data collection commenced the study was approved by the South West 5 Rec, UK National Health Service ethics committee (10/H0107/75, approval gained January 2011). For each hospital we obtained initial permission from the clinical lead for cardiology before approaching staff and teams in specific wards and departments. One hospital we approached declined to take part. As this was a study involving ward observations and staff shadowing of a clinical condition needing care in multiple hospital departments, it was accepted by our ethics committee that it would be impossible to gain written consent for all participants encountered. In agreement with our ethics committee we sought informed verbal assent from all participants being observed, followed by written consent of patients, staff and carers when the researcher observed their care more closely or requested an interview.

Data collection

Data collection was conducted in five different ways: (i) observation of care for a purposive sample of NSTEMI patients from admission through to discharge (n=68); (ii) observation of staff at each site (n=199); (iii) interviews with clinicians and managers (n=143); (iv) interviews with patients including the same patients 30 days after discharge (n=53); and (v) patients’ medical records (See Table 2). Fieldwork was carried out over two weeks at each hospital (June 2011 - August 2012), focusing on the processes of admission, diagnosis, treatment and discharge. The researcher, JH, had previous experience in conducting ethnographic interviews and observations in hospital settings, including research for her doctoral thesis. Staff identified and introduced likely ACS patients to the researcher early in the patient’s hospital stay and the researcher then regularly visited them and asked staff about their care. The study was described to participants as trying to identify the different approaches to NSTEMI care with the aim of sharing the best practices more widely. Assurances of strict confidentiality of any critical feedback was given to all participants and, for patients, that their care would not be affected by anything they said. The researcher was White British and did not have a clinical background but was passionate about improving patient care and the potential of qualitative research to achieve this. To minimise social desirability bias, staff behaviour and accounts of care were compared with observations and accounts of care from other staff and from patients. Patient participants were purposefully selected to represent diversity of age and gender (see Table 3), with carers sometimes being included in the interviews. Staff participants were selected to represent member identified categories and variation of roles within each hospital (see Table 4). We also paid special attention to specialist cardiac nurse roles as part of the theoretical sampling strategy. Interviews with staff and patients varied with some lasting a few minutes and others lasting over one hour. Verbal data were audio-recorded and transcribed; observational data were written as field notes. The transcripts and fieldnotes were not returned to participants or hospitals for comment or correction.

Data analysis

Data analysis aimed to identify variation in care, the possible reasons for observed differences and if and why it mattered. The interview topic guides and observations were informed by clinical guidelines available at the time of study (NICE, 2010, 2016; NICE 2010, 2013), existing literature on the quality of NSTEMI care as well as experiences at the two initial sites. We inductively identified categories from within the data, coded all of the data within these categories and used constant comparison to check items of coded data within each category for similarities and differences (Glaser & Strauss, 1967). The analytical approach followed many of the broad principles of grounded theory: simultaneous data collection and analysis; analytical codes developed from data not using preconceived logically deduced hypotheses; theoretical sampling; memo making as a bridge between coding and writing; and conducting literature reviews after developing an independent analysis. Our triangulation of data included both clinician and patient perspectives. Strong themes emerged on core areas which we felt gave us adequate information power (Malterud, Siersma, & Guassora, 2016) rather than data saturation as in such a complex clinical field, it is hard to claim full data saturation. The categories and code structure were developed by JH, HC and ME in collaboration with team members (KF, GF, CD and RJ). Half of the team had a clinical background (GF, RJ, CD, AT, HH) with specialisms in nursing (CD), cardiology (AT, GF, RJ) and general practice (GF, RJ) and the other half were non-clinicians with a variety of social science backgrounds (JH, HC, KF, ME).

Monitoring a wide group of patients to enable identification

Monitoring a wide group of patients as they enter the hospital system and ongoing outreach work with patients and staff in multiple locations across the hospitals was part of the NSTEMI challenge. Identifying, and then closely monitoring NSTEMI and potential NSTEMI patients, was fundamental to care.

Patients arriving in the Accident and Emergency department (A&E) are initially seen by generalist staff. NSTEMI diagnosis took time to emerge and, in many cases, required cardiac specialists to combine symptom interpretation and test results. As diagnosis was often complex and protracted, close monitoring in suspected cases was also very important. Having robust organisational arrangements whereby generalist staff (in places like A&E) could quickly, regularly and easily call on specialist cardiac staff for patients with symptoms like chest pain was essential for NSTEMI diagnosis but varied significantly across our hospital sample. Only half of the hospitals in the sample employed cardiac specialist nurses whose role focused on close monitoring to facilitate early identification of patients with NSTEMI (or possible NSTEMI). Other hospitals, for example, employed cardiac nurses whose main job was data reporting, or did not employ cardiac specialist nurses at all. In Hospital 3, the specialist nurse described how he undertook a huge variety of everyday monitoring related tasks and informal practices that might be needed in the support of identification in A&E:

Raised troponin levels helped to identify possible NSTEMI patients. However, troponin may also be raised for other reasons such as kidney disease, making diagnosis difficult. One patient, who had kidney disease (renal failure), described his experience:

Identifying possible NSTEMI patients who may have been missed at the time of admission or who had a cardiac event while in hospital required particular approaches to monitoring also involving specialist expertise. Most, but not all, hospitals in the sample used daily lists of troponin measurements as a method to try and identify NSTEMI and possible NSTEMI patients located around the hospital. Cardiac nurses in Hospital 8 described checking troponin lists daily as a laborious but necessary fail-safe method:

The danger of not having specialist cardiac staff closely accessible, available and checking over patients was that NSTEMIs could get easily overlooked. One senior cardiac nurse described in some detail how she persisted with her suspicions, monitoring and investigations on behalf of an NSTEMI patient in A&E, highlighting the complexity of diagnosis and frequent need for specialist interpretation:

Prioritisation of high and medium risk patients enabled by close monitoring

Ongoing close monitoring of NSTEMI or potential NSTEMI patients was required to prioritise patients at medium or higher risk of another cardiac event or, to identify patients who were clinically deteriorating. Again, good access to specialist cardiac support was often critical. For instance, the lack of specialist input in A&E may have meant that a patient in Hospital 9 was not prioritised appropriately and only received an intervention at the upper time limit recommended for interventions (96 hours maximum):

Where ongoing pain is reported, risk is increased.[16] The patient in this case was told the blockage in her artery was large. In an interview she explained that her doctor thought she should have received an intervention much earlier:

In Hospital 9 where this patient’s treatment was slow, cardiac ward staff reported an inadequate A&E triage system and poor access to specialist cardiac advice. With very little initial filtering of NSTEMI or possible NSTEMI patients, as patients entered the hospital a wide assortment of chest pain patients were sent to the cardiac wards with staff acknowledging that a high proportion of these were being inappropriately admitted, for example, with gastric rather than heart problems. In the following example, the cardiologist refers to system as being ‘crumpled’ due to poor filtering and prioritisation systems at the hospital entry point in A&E:

Hospital 5 stood out as having the most effective systems for NSTEMI patient triage and prioritisation. Part of this system included having an overnight cardiac registrar checking A&E and other wards for possible NSTEMI patients. While specialist cardiac support in the other hospitals was often reduced significantly at nights and weekends, in Hospital 5 the 24/7 integrated cardiac specialist approach meant that the overnight registrar helped to identify and prioritise NSTEMI patients and had a robust morning handover. Where all the other hospitals had daily ward rounds (or sometimes less frequently in the case of Hospital 4), Hospital 5 had two multidisciplinary ward rounds per day. In the following quote, a nurse explained how the overnight registrar doctor (or ‘reg’) duty system worked. The quote highlights how closely cardiac specialists worked with generalist staff and describes a clear process for double checking decisions with senior colleagues:

Certain categories of NSTEMI patients were not considered for prioritisation. Two hospitals automatically excluded patients over the age of 85 from cardiac wards, unless there were exceptional reasons. This conflicts with ESC guidance that older people benefit equally from intensive management (ESC 5.8 web addenda, Roffi et al., 2016; Kaura et al., 2020). For example, in Hospital 5 we traced the care of one 87year-old woman with a diagnosis of NSTEMI who stayed on a care of the elderly ward. On this ward she did not see a cardiologist, nor have her planned echocardiogram and did not benefit from the close cardiac monitoring necessary for the medical management of her condition. She died shortly after leaving hospital (P34, OB:FN and medical notes).

Close monitoring dependent on flexible catheter laboratory and bed management

Effective monitoring of an NSTEMI patient included being able to respond quickly and appropriately to changing patient needs. Making an accurate diagnosis with angiogram and, where appropriate treating disease with angioplasty, were both sometimes needed. As well as differences in the availability of specialist cardiac staff to support timely and appropriate responses, there were also considerable differences between hospitals in their ability to access catheter laboratory facilities (where angiograms and angioplasty were carried out), and in their bed management systems.

Guidelines recommends an intervention window of 96 hours.[14,15,16] However, the wide variation in the availability and number of diagnostic and treatment facilities limited hospitals’ ability to respond within that window. Some hospitals had no catheter laboratories (Hospital 2), some hospitals only had facilities and trained staff for angiogram diagnosis (Hospital 8) and some hospitals had capacity and staff for both angiograms and angioplasty (Hospitals 1,3,4,5,6,7,9,10 – see also Table 1). NSTEMI patients needing angiogram/angioplasty who were initially admitted to smaller district hospitals with limited or no interventional facilities, required transfer to a hospital which did have the facilities. For patients needing transfer, delays in treatment (or worse, no treatment) were regularly reported. In one smaller hospital, although they had catheter laboratories and offered both angiograms and angioplasty, half of the six patients observed went home without their planned angiograms (Hospital 4, OB:FN). In one hospital the cardiologist highlighted regular delays when transfers were needed between hospitals:

One patient described their frustration at having to wait in a smaller district hospital to be transferred to a larger hospital with better facilities:

Being admitted to a larger (‘tertiary’) hospital with all the facilities within the same hospital, however, did not always mean that required interventional work was available for NSTEMI patients. In the larger hospitals there was competition from STEMI patients who were always prioritised over NSTEMI patients in accessing a slot in the catheter laboratory.

Closely related to being able to secure a slot in a catheter laboratory, hospital bed management was essential to support patients with the appropriate level of cardiac care, especially after interventional treatment. Hospitals managed their beds in various ways and we identified several different methods that appeared to be effective. For example, in Hospital 7, bed meetings took place four times per day and it was compulsory for all staff to attend these meetings. In many hospitals however, ineffective bed management practices at times undermined patient care:

In Hospital 5 a bed management system had been devised that was led by nurses and was very flexible and placements were not dependent on the approval of doctors (which often slowed how soon a bed could be secured). Referred to in Hospital 5 as ‘patient flow’, one fieldwork diary entry captured this constant need to juggle bed spaces according to patient needs:

Close monitoring within an integrated system for timely interventional responses

Having a whole system approach facilitated responding in a timely manner. By this we mean the NSTEMI patient journey had been well thought through and there was a cumulative effect of having robust bed and catheter laboratory systems, underpinned by regular ward rounds as well as cardiac specialist input. For example, a patient case in Hospital 5 illustrates a seamless process for screening and swiftly treating patients. This patient was quickly found a bed in a specialist ward, his ongoing pain was under close and repeated review by specialist cardiac staff and there was a rapid interventional response:

By contrast, in Hospital 4, bed management challenges, lack of an integrated cardiac system, and the limited access for NSTEMI patients to the catheter laboratories hampered timely intervention. In Hospital 4 there was a particular problem around beds because they had dual condition wards where both cardiac and respiratory patients were placed. Having dual conditions in one ward meant that there was sometimes confusion over which patients needed monitoring for which conditions and ultimately the responsibility for individual patients. On occasion, cardiac doctors missed seeing their patients on ward rounds. For example, one 82-year-old NSTEMI patient stayed for nine days in Hospital 4 but went home without her planned angiogram and two important medications missing. The nursing staff said that she had been missed on a ward round and this patient confirmed that she had only seen the doctor once:

Underlying data

The underlying data are not publicly available due to their containing information that could compromise the privacy of research participants and due to data security restrictions.

The research was given ethical approval on the condition that participants confidentiality would be maintained. We assured our committee ‘the absolute anonymisation of all data will protect the privacy of all patients’. Given the nature of the work the only reasonable way to ensure this confidentiality for both patients and staff is to not make data publicly available and to exercise care and caution of data requests. For example, we consider the data related to individual patient care such as patient’s medical notes to be too identifiable and in interviews, critical reflections of management and hospital care were discussed. Due to the small number of hospitals involved in the study, staff could be easily identified by their roles as there was often only one matron or specialist nurse role per hospital. The MINAP data on hospital performance (30-day mortality after admission) is not openly published and we do not have permission to disclose this. In our application to the MINAP committee we stated the research team would ‘not publish the hospital’s identity in any way or publish any linked information about individual hospitals and their outcome group (e.g. good or poor)’. We were specifically asked by our NHS ethics committee not to divulge to the participants that the research was looking at poor performing hospitals as well as top performing ones. The anonymised data that support the findings of this study are, however, available on request to bona fide researchers and will be considered on a case-by-case basis. We will consider, for example, requests for anonymised data that supports particular claims made by the authors. Approvals for such requests will require the consideration of the study team’s ethics committee but may also require some reassurance from the requesting researcher’s ethics committee on their planned use of data. Requests to access the data should be directed to South West-Frenchay UK National Health Service ethics committee, previously known as South West 5 Rec (frenchay.rec@hra.nhs.uk).

Extended data

Zenodo: Topic guide and observation guide for Understanding Variation in Patient Care. https://doi.org/10.5281/zenodo.6360572 (Cramer et al., 2022).

This project contains the templates for the interview questions and observation guides used in this research.

Data are available under the terms of the Creative Commons Attribution 4.0 International license (CC-BY 4.0).