There is a common narrative that urgent and emergency care of older people with dementia is not optimal. There are also anecdotal stories that older people with dementia are more likely to be taken to hospital because ambulance crews cannot assess them properly or access alternative services. People may be calling 999 for ‘inappropriate’ reasons, such as non-medical crises and out-of-hours (OOH) services, and care homes may be unnecessarily calling ambulances. A recent review found the use of ambulance services or EMS by older people with dementia is not well understood. The literature reviewed demonstrated a concern for this group, need for training for EMS staff and awareness that current services are not optimal (Buswell et al, 2014). We set out to use aggregate data from ambulance service patient care records (PCRs) to understand 999 call-outs to older people (>75 years) with dementia, aware from the outset that dementia may not be consistently recorded.
Aims and objectives
We aimed to understand the use of emergency ambulance services by older people with dementia in two English counties.
Our study objectives were:
We looked at records for persons aged 75 years and over. While health services often define older persons from aged 65 years, those who are 75 years and over have more health needs than the recently retired (Barnett et al, 2012). This also kept the sample size manageable for this paper-based sample.
Methods
This initial study was carried out on a small sample of paper records, as it was not known in advance where in the patients' records dementia would be recorded. With the electronic PCRs (ePCR) for this region there are multiple possible places that dementia could be recorded and, as we were not confident we would find them all, an initial review of paper PCRs was needed to establish a baseline. We anticipated that dementia may not be well recorded because there is not a specific ‘tick box’ in the PCR to direct recording. However, in 2012 some dementia awareness training had taken place across East of England Ambulance Service NHS Trust (EEAS) and this may have affected how crews recognise and record dementia.
Non-patient-identifiable data were obtained from a review of four days' PCRs of EEAS 999 call-outs to patients aged ≥75 years in two counties (Cambridgeshire and Hertfordshire). The days were two midweek days and two weekend days in February, May, August and November in the financial year 2012–13. Fields from which we extracted data were:
The whole record was searched for mention of dementia and for text suggesting possible dementia. How dementia was documented by crews and where in the PCR it was noted was recorded.
The EEAS computer-aided dispatch (CAD) system provided the total number of call-outs across the East of England and for just Cambridgeshire and Hertfordshire for a) all ages and b) those aged ≥75 years.
The age-sex distribution of our sample was compared to the general population in Cambridgeshire and Hertfordshire, and dementia prevalence as recorded in the PCR was compared to prevalence estimates from the Cognitive Function and Ageing Study (CFAS II), a large UK-based longitudinal multicentre study looking at health and cognitive function in older people. (Matthews et al, 2013).
To determine whether it is possible to collect information from the PCR data on possible contextual factors that may influence outcomes for older people with dementia after 999 ambulance call-outs, information was collected on outcome, time of call, paramedic impression and social family history. All proportions are reported with confidence intervals.
Results
There were 7 922 CAD records for call-outs for all ages across the region on the four days, and 2 304 for Cambridgeshire and Hertfordshire. Of the CAD records, 4 243 (54%) were scanned copies of paper records. The scanning software could not read the age field in 650 (15%) of these. The remaining 3 593 paper-based records for the region were searched to select those for people aged ≥75 years: 1229 records. From these we extracted data for call-outs to people aged 75 years or older in Cambridgeshire and Hertfordshire only: 358 records.
Over a third of call-outs were to ≥75-year-olds
Call-outs to people aged ≥75 years accounted for 35.4% (95% CI 34.4–36.5) of all records, 34.2% (95% CI 32.7–35.8) of the available paper-based records and 36.0% (95% CI 34.1–38.0) of records for Cambridgeshire and Hertfordshire only. In Cambridgeshire and Hertfordshire, 7.8% of the general population are aged ≥75 years.
Age-sex distribution
In our sample 216 patients attended were female (60.3%, 95% CI 55.3–65.4) and 142 were male (39.7%, 95% CI 34.6–44.7). This is in line with the sex distribution for the Cambridgeshire and Hertfordshire ≥75-year-old population: 58.8% (95% CI 58.6–59.1) female and 41.2% (95% CI 40.9–41.4) male (Source: ONS Mid-2012 Population Estimate).
The age distribution of our sample suggests the ambulance service sees higher proportions of older old people (≥85-year-olds) than the general population distribution for Cambridgeshire and Hertfordshire (Source: ONS Mid-2012 Population Estimates): higher by a factor of 1.57 for 85–89 year-olds and 2.20 times higher for ≥90-year-olds (Figure 1). Conversely, proportions of EMS call-outs to not quite so old people are lower: 75 to 79-year-olds account for only 19% of call-outs, less than half (48%) the proportion of the local population this age (39.7%). EMS call-outs to 80 to 84-year-olds (27.9%) more closely reflect the population distribution (30.8%).

Living situation
The type of residence was indicated in the ‘social/family history’ field in 344 records, 96% of all 358 records. Of these, 54/344 (15.7%, 95% CI 11.9–19.5) were people in nursing or residential care homes and 290/344 (84.3%, 95% CI 80.5–88.1) were community-living.
Just over a third of calls-outs to those living in the community were to people recorded as living alone (38.6%, 95% CI 33.0–44.2), close to half of the women and a quarter of the men (respectively 44.7%, 95% CI 37.6–51.8 and 22.4%, 95% CI 15.1–29.7). This is lower than the 50% of ≥75-year-olds living alone nationally (women: 61%, men: 35%) (Source: 2011 General Lifestyle Survey, Office for National Statistics). Although crews recorded details about home situation and level and type of care, this was not sufficiently consistent to be used.
Dementia recording
Dementia was recorded on 14.5% (95% CI 10.9–18.2) of records and another 7.0% (95% CI 4.3–9.6) had details noted that suggested possible dementia or cognitive impairment such as ‘memory loss’, dementia drugs or often simply ‘dementia?’.
Of the definite dementia recording, 51/52 were recorded in the ‘previous medical’ field; the other was in the ‘history/MOI’ field. Of the 24 cases with notes indicating possible dementia, 18 were recorded in the ‘previous medical’ field, five in the ‘treatment, advice and notes’ field and one in the ‘history/MOI’ field.
Levels of dementia recording are in line with estimates from CFAS II (Matthews et al, 2013) by age, sex and residential status—living in care-homes or community-dwelling (Table 1) apart from care home residents: the proportions with PCR-recorded dementia were lower than would be expected in this population, significantly lower in the more numerous age-bands—women aged ≥85 years and men aged 75–84 years.
Age band | EEAS PCRs % (95% CI) | CFAS II % (95% CI) |
---|---|---|
Women | (not indicated = 6) | |
Care home | n =38 | n =130 |
75–84 | 62.5 (29.0–96.0) | 76.0 (59.3–87.2) |
85+ | 33.3 (16.5–50.2) | 71.1 (60.0–80.1) |
Community | n =172 | n =2 123 |
75–84 | 8.8 (2.6–14.9) | 5.9 (4.9–7.8) |
85+ | 13.0 (6.2–19.9) | 17.0 (14.2–21.9) |
Men | (not indicated = 8) | |
Care home | n =16 | n =42 |
75–84 | 14.3 (-11.6–40.2) | 72.0 (44.6–87.0) |
84+ | 33.3 (2.5–64.1) | 55.6 (31.5–77.4) |
Community | n =118 | n =1 612 |
75–84 | 10.9 (3.3–18.6) | 6.4 (5.2–8.1) |
85+ | 11.1 (2.7–19.5) | 11.9 (8.6–16.3) |
Call-out reasons, time of call and outcomes
The reason for a call (see Figure 2) was recorded in 97% of PCRs, most commonly a fall: 27.5% (95% CI 21.0–34.0) of this ≥75-year-old sample or 9.4% overall. This is in line with estimates that about 8% of emergency ambulance responses each year are for a ‘fall’ (Snooks et al, 2012). For those PCRs where dementia was recorded, almost half were for a fall: 24/50 (48.0%, 95% CI 34.2–61.8). This single problem far outweighed others, the next largest categories being less well-defined composite categories: 20.9% (95% CI 15.5–26.3) classed ‘other medical problem’, where a specific medical complaint was recorded, and 14.6% (95% CI 10.3–19.0) ‘non-specific’, a category which included various descriptions such as ‘off legs’, ‘unwell’, confusion or pain. These twelve categories that we derived for ‘paramedic impression’ required a level of clinical judgement and interpretation of the free text and are not validated. However, we are confident with the ‘fall’ category as it was always clear where a fall was documented. It was not possible from this routine data to be clear how many of the falls were from a standing height or how many represented someone being found on the floor.

There were no significant differences in the times of call-outs to people whose PCR recorded they had dementia and those with or without any suggestion in their record of possible dementia.
Just under two-thirds of ≥75-year-olds attended to by an ambulance crew were transported to hospital, regardless of age, gender or whether dementia was recorded: 217/351 (61.8%, 95% CI 56.7–66.9). It should be noted that we are aware of differences in rates of conveyance to hospital between outcomes recorded in paper versus electronic PCRs—higher transport rates reported in ePCRs, as we have reported elsewhere (Buswell et al, 2015).
Limitations
It is possible that there is bias in this sample examined through paper PCRs, as there may be differences in the populations for whom paper PCRs and ePCRs are used. Anecdotally, paramedics have reported that they are more likely to complete a paper PCR if they are leaving the patient at home and an ePCR if they are transporting a patient to hospital (Buswell et al, 2015). Therefore, there may be some bias in the transported rates reported and indeed the population characteristics.
The coding system used to categorise the free text description in the ‘presenting complaint’ field has not been validated. For some categories, such as a fall, the description is unambiguous and we report these reasons with confidence. Nonetheless, we can only report the paramedic impression, rather than the true reason for a fall. Whether a fall constitutes the presenting complaint or whether in fact it is the outcome of other underlying conditions or issues is open to debate. The high frequency of ‘falls’ in older people makes this is an important indication to report.
Discussion and conclusions
This record review quantifies (in a small and population-specific sample) what many ambulance crews may perceive in their day-to-day practice: that they are called to a significant number of older people; or one-third of call-outs are to those ≥75 years when only 7.8% of the Cambridgeshire and Hertfordshire population are ≥75 years. We had anticipated poor recording of dementia, so were encouraged by the level of recording of dementia (14.5%), a proportion in line with population-based age-specific prevalence estimates, perhaps reflecting increasing awareness of this group of patients. There are two possible explanations: that the population ambulance crews are called to reflect the general population of this age and recording is fairly accurate, or, if recording is poor, they are seeing higher levels of dementia and those recorded are just the ‘tip of the iceberg’. One might presume that crews record dementia only where it is recorded in notes on scene or a family member or carer alerts them to the diagnosis. However, recording was rather lower for care home residents, possibly suggesting ambulance crews and care home staff regard dementia as less note-worthy because it is more common in this setting. Ambulance crews are routinely careful not to diagnose and are taught to put ‘?’ after any clinical impressions they have. We decided to quantify the records where dementia or cognitive impairment was a possibility (a further 7% besides definitely recorded dementia) because wording such as ‘confusion—unclear whether more than usual’ clearly cannot be used to distinguish, say, delirium from delirium superimposed on dementia. Suggesting recording is ‘poor’ is not a reflection on the practice of crews. The records we reviewed had no clear field to record dementia if it was not the presenting complaint and therefore it was up to individual practitioners whether it was relevant and how to record it.
The high level of completion by crews of the social/family history field in forms was also noteworthy. Although recording in this free text field was not consistent, notes frequently detailed how the person was living: alone or with carers, and their care arrangements. This suggests that ambulance crews believe that this information is important to record for the ongoing care of the person they are attending.
The most common reason for a call-out was a fall, just over a quarter overall but almost half where demential was recorded. This review of routine records did not allow further analysis. Future falls research would do well to look at the impact of dementia.
The changing role of the ambulance service from a patient transport service to a service that brings care to the patient, responds to non-emergencies and has a role in promoting health has been discussed in documents and reports (Department of Health, 2005; 2007; Ambulance Service Network, 2010; National Audit Office, 2011) and in the recent NHS England Urgent and Emergency Care Review (NHS England 2013a; 2013b). With the national focus on dementia care (Department of Health, 2009) these findings provide a basis for future much-needed research on the use of ambulance services by people with dementia for urgent and emergency care. In particular, their role as a link in how care is co-ordinated for older people with dementia who use emergency ambulance services but may not need secondary care needs closer examination. From our experience using these paper-based PCRs, and from discussions about the possibilities of using the ePCRs in EEAS, these would not seem an ideal data source for studying use of ambulance services by older people with dementia. It may be different in other areas of the UK or internationally if more consistent record keeping is used.
Key Points
Research Governance: Letter of Access granting permission to carry out the study from EEAS and included in the RODES study protocol, Research into Older people with Dementia and their carers’ use of Emergency Ambulance Services (RODES), Protocol number: HSK/SF/UH/00049, Dr R Southern, University of Hertfordshire, Health and Human Sciences ECDA Chairman.
Conflict of interest: none declared