Introduction

Critical care is a term used to describe intensive care and high dependency care (HDC) [6, 15]. In the UK, children requiring paediatric intensive care (PIC) require support for one or more organ failures. PIC patients are usually cared for in designated intensive care units (ICUs) staffed by specialist medical and nursing personnel. Although PIC may be initiated in other settings such as the operating theatre or emergency care unit, usually PIC takes place in the ICU. The same cannot be said for HDC where HDC is not synonymous with a dedicated high dependency unit (HDU). Children requiring HDC may be very sick and require a high level of medical and nursing intervention but this level of care is provided by a diversity of healthcare personnel in many hospitals and hospital wards.

The lack of a clear definition makes it difficult to describe HDC. Few attempts have been made to quantify HDC in the UK possibly because of the loose criteria that allow individual healthcare professionals to subjectively assign patients to this level of complex care [4, 11, 24]. As a consequence of inadequate definitions and the heterogeneity of HDC in paediatric secondary and tertiary care, there is a lack of information on the various locations and volume of HDC delivered in the UK.

To inform service planning the Paediatric Critical Care Network of the Yorkshire and Humber Strategic Health Authority in West, North and East Yorkshire commissioned a project. The aim was to produce a working definition of HDC to allow the systematic assessment of the volume of HDC inpatient activity in hospitals and wards in the region. We report on the development of a measurement tool and its application to measure the volume of HDC by shift period, children and bed days, over time and by ward area.

Methods

A prospective cohort study collected inpatient activity data for children of all ages, assessed as receiving HDC [21]. Demographic and basic diagnostic information was collected for all children receiving HDC (Table 1). The 580 paediatric inpatient beds available in the region were spread between 36 hospital wards and 14 hospitals; these hospitals were administered by ten National Health Service (NHS) Trusts. National census data provides populations by 5-year age groups for children 0–14 years; the childhood population (0–14 years) of the study area in 2005 was 629,100 (7%) of the English population [17].

Table 1 Demographic and diagnostic criteria requested for children receiving high dependency care in West, North and East Yorkshire in 2005
Full size table

The measurement tool data collection form was piloted for the period of one month (2004) on 36 hospital wards aiming to establish the tool was capturing children receiving HDC [21]. Initially, the data collection form itemised fifty clinical interventions each derived from one or all of the following, therapeutic intervention scoring systems (TISS) [13], Department of Health guidelines [6, 7], interviews with nursing staff from each ward involved in the study, and regional multidisciplinary group workshops. Children were eligible for form completion if they had received one or more of the 50 interventions grouped under seven headings; 1; airway, 2; breathing, 3; circulation, 4; renal, 5; neurological, 6; IV fluids/infusions and 7; other. Analysis of over 5,000 forms submitted over a month period showed that the tool was capturing a proportion of children not deemed not to be receiving HDC by an expert group of consultants and nurses; during this exercise, the experts were blind to any demographic information or the hospital of treatment. As a result, the number of clinical interventions on the form was reduced to 42 for the main study data collection exercise where the additional outcome measure was added for the staff completing the form comprising the question ‘in your opinion is this child in need of HDC’?

The main study collected data from January 1st to December 31st 2005 from 36 hospital wards in four types of tertiary and secondary services as described by the Department of Health [6, 7]:

  • Tertiary specialist wards.

  • PICUs in tertiary hospitals.

  • General wards at District General Hospitals (DGHs).

  • Wards at a major acute general hospital.

The tertiary and secondary services were configured as follows: 14 specialist wards at two tertiary hospitals, and one regional burns unit located at a DGH provided HDC to support tertiary specialties. Three PICUs with 17 beds served the 15 specialist hospital wards and the 11 DGHs. Only six designated HDC beds were available, two on one tertiary specialist ward and four on a High Dependency Unit (HDU) at the major acute general hospital. All hospitals were expected to stabilise, intubate and ventilate a critically ill child and provide short-term PIC until the retrieval team arrived to transport the child to one of the three PICUs in two tertiary hospitals, hence the inclusion of the criteria ‘intubation and connection to a mechanical ventilator’. Acutely ill children receiving mechanical ventilation on the PICUs were excluded because the definition of PIC ‘intubation and subsequent care of the ventilated child’ was set by the Department of Health [6]. However, those children in a PICU not receiving mechanical ventilation but requiring ‘greater observation and monitoring’, the UK definition for HDC [6] for example children with severe asthma, septic shock and diabetic ketoacidosis, were included.

As in the pilot, data were captured by bedside nurses for any child receiving one or more of the itemised 42 interventions delivered during two shift periods (day and night shift) every 24 h. Guidance notes with severity parameters accompanied the data collection forms to prevent nurses from using subjective judgement (a copy can be obtained from the lead author).

Data was also collected by nurses on each ward for daily ward admissions, discharges and transfers in May and November. Data quality was continuously monitored with systematic checks for missing, out-of-range data values and inconsistencies [2]. Monthly data quality reports were returned to each hospital ward to facilitate a constant review of the data by nursing staff. Hospital visits were made to assess data accuracy by checking a random sample of patient data collection forms against case notes. Data were analysed in SPSS V15 [22].

Complex analysis of concordance/agreement between the intervention data (both for individual items and their multiple combinations) and the outcome measure for over 12,000 forms (50% of dataset) demonstrated that removal of a further six individual interventions improved levels of agreement [21]. Validating the reduced 36-item tool (Fig. 1) on the remaining independent half of the dataset (over 12,000 forms) found moderate levels of agreement between the tool and the outcome measure (nursing opinion) using the Kappa statistic [3].

Thus the final HDC measurement tool comprised 36 clinical interventions which might occur singly or in various combinations (Fig. 1). Each mutually exclusive intervention was weighted according to its level of dependency and listed in three hierarchical groupings labelled A, B and C on the data collection form. The 14 interventions in group A were assigned a score of six; the 15 interventions in group B were assigned a score of four and the seven interventions in group C, a score of two. To score whether a child required HDC for a defined shift period a nurse completed the form and added the scores. A child scoring six or more was assessed as requiring HDC, less than six, usual ward care. This final tool was applied to the data collected in the main study.

Fig. 1
figure 1

The high dependency care measurement tool

Full size image

Ethics approval was obtained from the Riverside London Ethics Committee and all Local Research and Development Departments from each of the 10 NHS Hospital Trusts. The Patient Information Advisory Group Approval now the National Information Governance Board [16] granted permission for the collection of individual identifiable data.

Results

The 36-item measurement tool captured 9,077 shift periods of HDC, corresponding to 4,538 bed days for 1,763 children of all ages within the Yorkshire region (Table 2). During the study period 1,433/629,100 (0.2%) children aged 0–14 with West, North and East Yorkshire postcodes received 3,325 bed days of HDC. The majority (n = 967, 55%) of children requiring HDC were male. Table 2 shows the breakdown of children requiring HDC by age.

Table 2 Hospitals, wards, shift periods, length of stay and sores for children receiving high dependency care by tertiary and secondary service in West, North and East Yorkshire in 2005
Full size table

By hospital ward, more (1,466, 16%) HDC was delivered on one of three PICUs than on any other hospital ward. Slightly more HDC was provided for the day shift (n = 4,627; 51%). There was a continual demand for HDC throughout the year, November was the busiest month for HDC (n = 897, 10%) and June was the least busy (n = 623, 7%). Just over one quarter of children requiring HDC were admitted with a surgical problem. Scores obtained using the HDC measurement tool are broken down in Table 2. Although the following three interventions occurred most commonly for children requiring HDC, they occurred in combination with one or more interventions because singly they did not score 6 (Table 3); (i) complex hourly fluid balance, (ii) continuous IV drug infusion other than analgesia and inotropes and (iii) pain requiring epidural or intravenous analgesia. HDC defined by a single intervention, i.e. an intervention with a maximum score of 6 accounted for 877 (10%) shift periods.

Table 3 Type, number and scores for the 36 clinical interventions using the HDC measurement tool (per shift period) by tertiary and secondary service in West, North and East Yorkshire in 2005
Full size table

Tertiary specialist wards

Of the 3,805 shift periods of HDC, 751 (20%) were for children from outside of the region. Just over one quarter of children were in the 10–15 age group (Table 2) and 299 (44%) children were admitted with a surgical problem. These hospital wards experienced large volumes of activity all year with peaks in HDC in February, April, August and November (Fig. 2). Forty percent of shift periods received an HDC score of 6 (Table 2). Invasive pressure monitoring (arterial and central venous) was performed on two wards, the burns unit and the ward with two designated HDC beds (Table 2).

Fig. 2
figure 2

Shift periods of HDC by month and hospital type in West, North and East Yorkshire in 2005

Full size image

A total of 346 children required analgesia during 2,108 (69%) shift periods of HDC. Children most commonly admitted to the regional burns unit for HDC were aged 1 to 4 years (n = 16, 36%), one quarter (n = 11) of which were for children with a Leeds postcode.

High dependency care in the paediatric intensive care units in tertiary hospitals

PICUs at the lead centre provided 2,745 (30%) periods of HDC for 608 (34%) children. PICUs accounted for 44% (n = 2,745) of all HDC provided in the tertiary centre (Table 2). In the PICUs, 57% (n = 344) of children were male and 30% (n = 181) were infants. Admission following cardiac surgery was the reason for more children (181, 30%) receiving HDC. HDC scores ranged from 6–72 using the HDC measurement tool, 11% (n = 297) of shift periods scored 6 and 68% scored 16 or less. Details of interventions performed for HDC frequently in the PICU are found in Table 2. Of the shift periods of HDC for children requiring stable long-term domiciliary ventilation (LTV) with a tracheostomy, 693 (89%) shift periods or 347 bed days for 16 children were provided on one PICU. Over 14 children were responsible for 37% of the total LTV activity on PICU.

Of the 851 admissions to the three PICUs, 34% (n = 290) of children were never intubated; this varied between the three PICUs from 19–60%, these children required observation and intervention other than intubation. During 2005, 151 children were refused admission to one of the three PICUs because the units were full to capacity.

High dependency care on the general wards at district general hospitals

Eleven DGHs and 14 hospital wards provided HDC for 1,866 (21%) shift periods for 589 (33%) children. The majority of children were male (323, 55%) and the largest group were infants (n = 157, 27%). Nearly a quarter of children were admitted with a respiratory illness/croup (n = 134, 23%). There was a peak in activity during October, November and December (Fig. 2). Scores for HDC ranged from 6 to 62, half (n = 941, 50%) of which scored 6 (Table 2). HDC for children receiving stable LTV was provided for 77 (10%) of all regional shift periods of HDC for nine children at six DGHs (Table 2). Invasive pressure monitoring was only provided for children receiving PIC stabilisation before transfer to a PICU (Table 2). Of the 2,690 children admitted to a DGH in May, and 2,718 in November, 8% (n = 215) and 12% (n = 326) required HDC, respectively.

Wards at the major acute general hospital

The majority (99, 60%) of children admitted for HDC were male and one quarter were infants. Surgical problem was the most common (n = 45, 27%) reason for children receiving HDC throughout the hospital. Within the HDU, 47% (n = 236) of shift periods scored 8 or less (Table 2). HDC for LTV was provided on the HDU (Table 2). Nearly one quarter (n = 156, 24%) of all HDC shift periods were outside of the dedicated HDU provided at this hospital where 60% (n = 31) of children received HDC for a surgical problem.

Discussion

New information on the volume of paediatric HDC delivered in a demographically representative region of England is presented. Our empirical study collected a comprehensive dataset from children admitted to four different types of hospital representing all major specialties. The results report the level of HDC occurring on all ward types and clearly demonstrate the delivery of HDC was not restricted to any specific location of care and took place in varying amounts on all wards in the region.

Assessment of HDC activity in DGHs in May and November found that 8% and 12%, respectively, of all children required HDC, a level consistent with the English Department of Health [7] who estimated that 5–15% of children admitted to a DGH required HDC. PIC is well-described nationally [18] and includes information on HDC delivered in PICU. In 2009, it was reported that 76 (24%) of PIC beds were designated for HDC in 17 (53%) PICUs, an increase from 33 (13%) beds in 13 (42%) PICUs in 2005 [18, 19]. Despite the proportion of designated HDC beds, 28% of admissions to PICU did not require invasive ventilation (associated with PIC). Evidence from PIC and our study confirms that HDC is frequently delivered on PICUs. The centralisation of paediatric intensive care delivered in designated beds and units has improved outcomes for critically ill children [20]. Evidence for similar successes for HDC appears to be lacking, potentially as a consequence of the absence of clear definitions and the diverse locations of care. This study has taken the first step in attempting to address the issue of centralisation in HDC by investigating the feasibility of definitions and quantifying and describing the settings where HDC is delivered. This provides a stepping stone for future studies examining potential benefits of centralised care. In the UK, a Paediatric Critical Care Minimum Data Set [10] of clinical interventions is being developed by a team of medical and nursing experts using information from this study to gather details of each child’s admission to provide a systematic basis for economic costing; ultimately, these data will reflect the true identity of HDC occurring in PICU and hospital wards nationally. It is possible therefore that payment for HDC in the future will only be made for delivery of HDC in a centralised location where staff have the skills to care for such children.

Although the location of provision of HDC is dependent upon the local service model, the overall numbers of children requiring HDC is independent of location. The study was commissioned in the Yorkshire region by a proactive clinical network with a view to provision of results that could be employed both locally and nationally. The Yorkshire region is representative of England in terms of population demography and socioeconomic status [9]; therefore, assuming that healthcare services and inpatient profiles are also equivalent, our results are generalisable to England. Applying our results to a national population, the requirement for children aged 0–14 years would be 63,434 national HDC inpatient bed days.

There were variations in the volume of HDC by age and between hospital wards. Although the large amount of HDC provided for infants possibly reflected health problems associated with immature physiological development, the majority of HDC was provided by the tertiary specialist wards, where more HDC was delivered to children aged 10–15 years and where 20% percent of the total shift periods of HDC were for children from outside of the region, which reflected the sheer volume of complex work. One possible explanation for the different age profiles between the tertiary specialist wards compared to other settings was their case mix, including young people with complex, chronic conditions which required HDC level of care often repeatedly during frequent admissions. Specific factors contributing to children triggering the HDC criteria on the tertiary specialist wards (Table 3) include:

  • the use of post operative intravenous/epidural analgesia in surgical patients;

  • delivery of inotropic agents prescribed to children on the cardiac/cardiology ward;

  • chest drain usage for both general and cardiac surgical patients;

  • complex fluid balance (measurement of three fluids in and two bodily fluids out) particularly in nephrology, haematology, oncology and hepatology patients;

  • continuous intravenous drug infusion (excluding analgesia and inotropic support) in nephrology, haematology, oncology and hepatology patients;

  • multiple intravenous drug boluses (four drug boluses in 12 h) in nephrology, haematology, oncology and hepatology patients.

Admission criteria for PIC vary across the UK; however, the three PICUs in Yorkshire delivered surgical HDC for children aged 0–16 years. The absence of dedicated HDC capacity in the tertiary centres resulted in the use of PIC beds for HDC. In addition, the lack of an LTV facility where children could be nursed with a domiciliary ventilator and associated tracheostomy also compelled children to stay on the PICU; both factors displaced acute admissions from the PICU and contributed to a high refusal rate. Although children were entered into the study up to the point of intubation (intubated children were not included), 2005 was representative of other PIC years (2003–2009) for admission rates (825–921, mean 871) and refusals (109–161, mean 130); therefore, our data expose well-recognised deficits in local provision.

The prospective cohort method was chosen to assess the incidence of HDC over a specified period of time and to capture a large study population over a wide geographical area [5]. The study sample was non-random but geographically defined and included all hospitals and wards where children were admitted and, therefore, potentially captured a total population. Owing to the heterogeneous locations across a large geographical region, institutional selection bias was reduced making the findings potentially generalisable to other geographical areas of service delivery. One limitation which may have contributed to a systematic bias in the data collected was the inability of the researcher to check the number of shift periods and children potentially missed during the data collection period. The consequences of this were a possible underestimation of the population requiring HDC. However, it was unlikely that a substantial underestimation occurred because of the large quantity of data returned each month over the year with little fluctuation. Collection of data was a nursing responsibility for which nurses were enthusiastically supportive for a measurement tool to ultimately assess their unrecognised HDC workload.

Although systematic biases are not easy to detect [12, 14], potential systematic biases may have been introduced by senior nurses underestimating HDC and junior nurses overestimating HDC. It would not have been practical to limit completion of the forms to one level of nurse only as junior on one ward may equate to senior on another ward. For future studies tests of consistency of opinion should be performed across locations and between nurses to inform of potential systematic biases.

Establishing the criteria to identify HDC was hampered by a lack of available literature, but this was offset by the knowledge of clinical staff and recommendations from the DH [6, 7] which were exploited to produce data collection forms and the subsequent measurement tool. The only practical way to determine if the period of care was HDC was to ask nurses at the time and although the level of agreement between nursing opinion and the tool represented 50% (overall) of the potentially achievable agreement beyond chance or ‘moderate agreement’, this level has been generally acceptable in this type of research [8].

Although the receiver operator characteristic method is one approach to establish optimal cut points, no gold standard or equivalent tool existed in the UK for external reference or validation with which to assist in the calculation of the sensitivity and specificity for each cut point [1, 23]. The final tool was underpinned by limited published data at the time of its development and evaluation [23] and therefore employed a weighted scoring system to accommodate for the opinions of nurses and clinicians involved in its development who agreed that HDC could be defined by a single intervention or various combinations of multiple interventions. The assignment of a value (2, 4 or 6) to each intervention was a pragmatic process which provided good discrimination between the three categories and produced scores which were clear and understandable to the assessors.

Conclusion

Without detailed information from national registries it is difficult to compare HDC services between networks in the UK and further between countries. Systematic information concerning need, appropriateness, skills of staff and methods for costing paediatric critical care has the potential to assist with evaluation of care and to differentiate between need and demand across all European countries. This is the first UK study to quantify the volume of HDC from within a clinical network and may be applicable to other counties because although the configuration of services has developed with minimal support from dedicated HDC beds and HDUs and may not be typical of services in other regions, it has developed a starting point for understanding how and where children present and remain for their period of HDC. From this study, essential information on the volume of HDC being delivered across the region was provided to commissioners, hospital managers and clinical staff and has provided a base for organisational and strategic planning of HDC. With PIC established, the definition of HDC and where care is delivered must become clearer to enable the sickest children to gain timely access to the correct level of care provision.