Objectives Internationally there is no consensus on defining and funding of paediatric high dependency care (HDC). This study tested whether a new UK Healthcare Resource Group (HRG) classification for HDC, with two categories of basic and advanced HDC, can identify children who consume greater staff resource. It also explored the impact of a change in basic HDC HRG criteria introduced in April 2011.
Design Observational study of medical and nursing staff resource use.
Setting 16 paediatric wards across 6 regional hospitals; 1 tertiary children's hospital (November 2010 to March 2011).
Participants 1098 infants and children admitted to paediatric wards.
Main outcome measures Number of children meeting criteria for basic and advanced HDC HRGs; care in a cubicle; medical and nursing staff costs, extrapolated from time spent at patient bedside.
Results 223 (20.3%) children met original HDC criteria (15.9% basic, 4.4% advanced). This fell to 88 (8.0%) with the change in basic HDC definition (3.6% basic, 4.4% advanced). Children who met original HDC criteria consumed greater bedside staff resource than those not meeting criteria (cost ratio 1.0:1.75:2.96 (non-HDC:basic HDC:advanced HDC)), with revised criteria identifying a (smaller) basic group with greater staff resource use (cost ratio 1.0:2.35:2.76). Being cared for in a cubicle was not associated with greater staff costs.
Conclusions HDC HRG criteria identify children who consume significantly greater staff resources. Revision of the definition has resulted in a large reduction of cases meeting the criteria but identifies a group consuming greater staff resources.
- Nursing Care
- Data Collection
- Health services research
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
What is already known on this topic
Historically a number of different tools have been used to describe children who require high dependency care (HDC).
A new definition of paediatric HDC has recently been introduced to inform Healthcare Resource Groups (HRGs) and Payment by Results (PbR).
None of these HDC tools or definitions has been validated.
What this study adds
HDC HRGs do identify children who consume greater staff resources—a separate tariff for patients who meet HDC HRG criteria is therefore justified.
A recent revision in the HRG definition of HDU reduces the number of eligible patients by 75% but identifies a group consuming higher staff resources.
Children looked after in a cubicle do not consume greater staff resources, questioning the HRG rule which uplifts HRG category in this situation.
High dependency care (HDC) describes the patient requiring close observation and monitoring but not intensive care. Previous estimates suggest that 5–15% of children admitted to UK hospitals require a period of HDC.1 In the UK, children's HDC is often delivered within ‘normal’ paediatric wards rather than high dependency units (HDUs) or paediatric critical care units. HDC is acknowledged to be an under-researched area.2 ,3
Internationally there is great diversity in critical care provision4 and in funding models for paediatric critical care.5–7 Many countries are grappling with defining and resourcing HDC (or ‘intermediate care’ as it is sometimes referred to8). There are concerns that standard ward funding is insufficient to cover extra HDC costs, but that tariffs associated with more complex intensive care are too high.8–11
In England, as in a number of other settings, hospitals are funded on the basis of activity12 with payments made using a system of codes, called Healthcare Resource Groups (HRGs).13 ,14 Every child admitted to hospital will generate a number of HRG codes based on their diagnosis/diagnoses. For example, a child with asthma will generate a core HRG code for acute asthma which results in one hospital being paid the same amount as another hospital admitting a child with acute asthma. This payment is the same whatever the child's length of hospital stay. Recognising that certain aspects of care add significantly to hospital costs, however, these are allowed to generate additional (‘unbundled’) HRG codes over and above the main diagnosis code.5 ,14 Critical care HRGs are an example, with one paediatric critical care HRG (bringing additional funding at a specified level) being added for each day that a child receives high dependency or intensive care. This additional funding is intended to compensate for extra resource use by these patients, with staff resources making up >80% of costs for looking after a critically ill child.15
Patients are allocated critical care HRGs using a classification based on the critical care interventions a child receives on any one day. The full list of relevant interventions is contained within the Paediatric Critical Care Minimum Dataset (PCCMDS) (box 1).16 Twenty-one items map to a hierarchy of seven paediatric critical care HRG levels, the two lowest being for basic and advanced HDC. In April 2011, the national definition for basic HDC was modified to raise the threshold for basic HDC. The key change was a requirement for combined ‘oxygen therapy plus pulse oximetry plus ECG monitoring’, compared to ‘oxygen therapy plus pulse oximetry’ or ‘ECG monitoring’ alone (table 1). It is not clear, however, whether either set of criteria does in practice identify patients with increased resource use.
Box 1 List of Paediatric Critical Care Minimum Dataset (PCCMDS) interventions and additional candidate interventions collected.
Description of activity
Airway and breathing
1. Apnoea requiring intervention in past 24 h
2. Upper airway obstruction requiring nebulised adrenaline
3. Severe asthma requiring intravenous bronchodilator or continuous nebs
4. Nasopharyngeal airway in situ
5. Care of a new tracheostomy (<7 days since tracheostomy formed)
6. Care of an existing tracheostomy (>7 days old)
7. Oxygen therapy 22–50% or <2 L/min
8. Oxygen therapy >50% or >2 L/min
9. Continuous pulse oximetry
10. Invasive ventilation via a tracheostomy (including CPAP)
11. Acute non-invasive ventilation (including CPAP)
12. Chronic non-invasive ventilation (including CPAP)
13. Continuous ECG monitoring
14. Central venous pressure (CVP) monitoring
15. Arterial line monitoring
16. Vasoactive infusion (inotrope, vasodilator, prostaglandin)
17. Cardiopulmonary resuscitation (CPR) in last 24 h
18. Arrhythmia requiring intravenous antiarrhythmic therapy
19. Temporary cardiac pacing
20. Intravenous thrombolysis (tissue Plasminogen Activator (tPA), streptokinase)
21. Diabetic ketoacidosis (DKA) requiring continuous insulin infusion
22. Intracranial pressure (ICP) monitoring OR external ventricular drain
23. Epidural infusion
24. Seizures requiring intravenous anticonvulsant therapy
25. Hourly (or more) GCS monitoring AND GCS <15
26. 20–80 mL/kg volume boluses in past 24 h
27. >80 mL/kg volume boluses in past 24 h
28. Parental nutrition
29. Haemofiltration OR Acute haemodialysis OR Acute peritoneal dialysis
30. Plasma filtration
31. Exchange transfusion
32. Extracorporeal liver support
33. Major surgery in past 24 h
34. Chest drain(s)
35. Number of prescribed IV drugs
CPAP: continuous positive airway pressure; GCS, Glasgow Coma Scale.
This study was designed to (i) investigate the medical and nursing staff resource use associated with children meeting definitions of basic (original and revised definitions) and advanced HDC, compared to children not meeting these definitions; (ii) investigate the medical and nursing staff resource use associated with additional interventions that have been put forward to the HRG Expert Working Group as being potentially worthy of inclusion within the PCCMDS and (iii) determine the interventions that are the main cost drivers for HDC.
The study was a prospective observational study undertaken between November 2010 and March 2011 across the West Midlands, UK. It included (i) a specialist tertiary children's hospital (six subspecialty areas: medical, surgery/neurosurgery, liver, cardiac, renal and haematology/oncology) with ward areas chosen to include neonates, infants and older children and (ii) a paediatric ward area at each of six regional hospitals, selected for highest patient acuity where more than one ward was available. Each ward was monitored, across a range of shifts, on six occasions separated by at least 1 week. Each monitoring episode lasted 6 h. In total 72 monitoring episodes were completed across 16 ward areas in 7 hospitals. The study was reviewed by the Research Departments who considered that ethics approval was not required as no patient identifiable information was collected. Staff, parents and children were given study information prior to observation.
Staff resource use
One observer recorded the number and grade of medical and nursing staff present at each bedside every 5 min, moving from bed to bed on a sequential basis. This method was used previously in an intensive care study.15 Time-points associated with ward rounds were identified and excluded, to avoid potential bias, given that patients admitted under a different medical team would not be seen by the ward round team. Allied health professionals were excluded as patients could require prolonged input for reasons unrelated to acuity of illness, for example, leg fractures requiring physiotherapy for rehabilitation.
A pilot study comprising two 3-h observation periods, validated the method against continuous monitoring of staff present at each bedspace, which would be expected to be the most accurate method but which has very high costs because of the need to observe all patients continuously. In the initial pilot 10-min observations were made but these did not align well with continuous monitoring (r2=0.70; data not shown). Much better agreement with continuous monitoring was found in the second pilot using 5-min intervals (r2=0.96; data not shown).
The 15 observers had health service backgrounds (including medical students, nursing students and healthcare assistants) and were trained in a 2-h classroom training session by a single study co-ordinator who also monitored data collection, shadowing each observer as they completed their first observation session and scrutinising each observation form and raising any data queries.
Minimum data set interventions
Information about PCCMDS and additional interventions that have been put forward as being potentially worthy of inclusion within the PCCMDS—referred to throughout the paper as ‘candidate interventions’—was collected from the senior nurse on duty at the end of each observation period for the period of observation. Individual interventions were mapped to basic (original and revised) and advanced HDC HRGs (table 1).
Valuation of resource use
Unit costs for each staff category were obtained from Unit Costs of Health and Social Care19 and the NHS careers website20 and are given in 2009/10 prices. For each staff category, a cost per 5 min of patient contact was estimated and used to calculate total mean bedside staff cost per patient. This was adjusted to take account of differences in number of observations per patient and then extrapolated to form a 24-h cost per patient. To do this, mean costs associated with each shift (early, late and night) were estimated for the total population and used to generate a cost index or weighting between shifts. This index was then used to estimate costs associated with unobserved shifts. For example, for patients observed in the early shift, corresponding late and night shift costs were estimated by multiplying the early shift cost by the corresponding index for early:late shift and early:night shift (see online appendix 1).
Patients were classified into three groups: non-HDC, basic HDC and advanced HDC. Differences in mean costs between these groups were tested (Kruskal–Wallis test)21 for original and revised definitions. Differences in costs across shifts and for cubicle versus open ward were also tested (Kruskal–Wallis).
Stepwise regression was used to determine which PCCMDS and additional candidate interventions had significant (p=0.05) positive relationships with total cost.
Data were analysed using MS Excel and STATA V.12.
Patient characteristics and overall costs
Seventy-two monitoring episodes were completed across 16 ward areas in seven hospitals, with 1098 neonates, infants and children observed (548 tertiary; 550 regional). Mean patient age (SD) across all (tertiary; regional) hospitals was 4.4 years (4.96) (4.2 years (4.39); 4.7 years (5.48)); 353 early shift patients were observed, 408 late shift and 337 night shift (table 2).
The estimated 24-h bedside staff cost across all patients and all hospitals was £101.29. There was a small statistically significant difference (p<0.01) between costs for regional (£100.06) and tertiary (£102.53) hospital patients. Across all hospitals, the early shift was associated with highest and the night shift with lowest costs (table 2).
Original basic high dependency care definition
Within the specialist hospital, 83 (15.2%) children met the definition for original basic HDC (range 3.1%–42.9% across wards); 28 (5.1%) children for advanced HDC (0%–25.0%) (figure 1A). Across the regional hospitals, 92 (16.8%) children met the basic HDC definition (range 6.6%–41.3% across hospitals); 20 (3.6%) met the advanced definition (0.9%–6.5%) (figure 1B).
Across all hospitals, patients meeting basic HDC criteria had significantly higher bedside staff costs than those who did not. Advanced HDC had a higher cost than basic HDC patients (table 3), with a relative cost ratio of 1.00:1.75:2.96 (non:basic:advanced). Similar results were obtained for tertiary and regional hospitals considered separately.
One hundred and four of the 223 patients (46.6%) who met HDC criteria were nursed in a cubicle. Bedside staff costs associated with being treated in a cubicle were not increased for either basic or advanced HDC criteria; in fact there was a trend towards reduced costs (table 4).
Revised high dependency care basic definition
Within the tertiary hospital the number of children who met the revised, tightened, definition for basic HDC was 17 (3.1%; range 0%–12.5% across wards). Across the regional hospitals this number was 23 (4.2%; range 1.3%–9.8% across hospitals).
Bedside costs associated with basic and advanced HDC patients were much closer with the revised definition, with a ratio of costs of 1.00:2.35:2.76 (non:basic:advanced).
Thirty-two (36%) of 88 children who met revised HDC HRG criteria were treated in cubicles. Cubicle use was associated with lower bedside staff costs, regardless of HDC classification (p<0.05) (table 4).
Impact of the change in high dependency care basic definition
Overall 135 patients no longer met basic HDC criteria with the change in definition (figure 1). Within the tertiary hospital the number of children who met the revised definition for basic HDC fell by 79.5% (range 50%–100% across wards). Across the regional hospitals this number fell by 75.0% (range 41.6%–85.7% across hospitals).
This fall in number of patients was offset by a rise in the 24-h bedside staff costs for basic HDC by approximately 43%, from £147.18 (original) to £211.21 (revised) (table 3).
Relationship between minimum data set interventions, additional candidate interventions and cost
Using stepwise regression, eight interventions were significant positive predictors of total bedside staff cost: continuous ECG monitoring, temporary cardiac pacing, acute non-invasive ventilation, oxygen therapy above 50%, arrhythmia requiring IV antiarrhythmic therapy, invasive ventilation via a tracheostomy, GCS monitoring and prescription of one or more intravenous drugs (table 5). Further exploration of the relationship between the number of prescribed intravenous medications and staff costs failed to confirm a significant relationship (data not shown).
Statement of principal findings
This study is the first to explore bedside staff costs associated with children meeting HDC criteria. It suggests that the PCCMDS classification has validity in relation to costs, distinguishing patients who are more dependent and thus require more staff resource. The revised basic HDC definition results in a marked reduction in numbers of children meeting HDC criteria, to around 8% of the inpatient population included in this study. Compared to the original definition, the revision identifies children who consume significantly greater staff resource, suggesting that it has ‘raised the bar’ in selecting a more dependent group of children; a group not dissimilar to the advanced HDC group in terms of resource use. In contrast to expectations, cubicle use was not associated with higher resource use.
Strengths and weaknesses of the study
A major strength of the study is that it provides comparable resource use estimates for patients not meeting HDC criteria and in both HDC categories, from different hospitals and across different shifts. It also includes patients receiving HDC in ward areas that are not directly commissioned to provide such care, thus providing widely generalisable findings. There are, however, limitations. First, despite the large overall sample size, data for some interventions are limited. Second, an implicit assumption is that staff time spent with patients was not limited by major resource constraints within the hospital, and thus, that the staff time spent with patients does not underestimate the time actually needed for care. Third, the assumption that observing a staff member at the bedside relates to a 5-min cost is somewhat crude. More detailed monitoring would have ensured more accurate estimates and enabled estimates of patient care (such as preparation of drugs) not undertaken at the bedside. This was not, however, feasible within available resources without seriously reducing sample size; further, good agreement was shown between the 5-min observations and more detailed monitoring.
The purpose of this study was not to identify the proportion of children meeting HDC criteria and caution would be needed in extrapolating from the proportion of children meeting HDC criteria in this study to the national picture. The choice of the hospital ward with the highest patient acuity and monitoring during the winter season would tend to overestimate the picture, while limiting data capture to a period of less than 24 h could underestimate the true proportion of children meeting HDC criteria on any one day.
This study has focused on medical and nursing staff time spent at the bedside. The costs described cannot be used to set a payment or tariff level for HDC as no attempt has been made to define total costs associated with providing care for a critically ill child. Alternative systems, such as the NHS Reference Costs process,22 will need to be used to estimate total costs and set payment tariffs.
Strengths and weaknesses in relation to other studies
While there is an extensive literature on payment systems in healthcare, including literature examining the adequacy of HRG methods,23 and the choice of funding algorithms,24 this has seldom been applied to paediatric critical care.16 Indeed, paediatric HDC has generated only a limited academic literature, with acknowledgement that this area requires further study.3 Previous UK audits25 ,26 have characterised numbers of children meeting HDC criteria, though each has used a different definition and no previous study has included the national definitions. Rushforth et al25 reported that 8–12% of children in hospitals across Yorkshire required HDC, while an audit of HDC across Scotland26 found a median duration of each HDC episode of 16 h (range 20 min to 316 days). The study reported here, however, provides the first attempt to explore the value of unbundled HRGs for the paediatric high dependency (or intermediate) care setting.
Meaning of the study
Important policy implications can be drawn from these findings, if it is assumed that differences in bedside costs represent costs more generally. These data suggest clear differences in staff costs for HDC versus non-HDC patients. Using the original basic HDC definition there was a clear gradient in cost differences. The change in basic HDC definition has the effect of reducing overall HDC expenditure by greatly reducing numbers of eligible patients while identifying those consuming greater staff resources. The revised definition results in less distinction in costs between basic and advanced HDC, questioning the need for two categories of HDC, at least from a cost perspective. The study provides evidence to support the inclusion of a number of additional candidate interventions to the PCCMDS.
The data also suggest that no step-up in HRG category is justified for cubicle care. This is perhaps surprising, given that paediatric intensive care standards in the UK recommend a higher nurse:patient staffing ratio if a child is looked after in a cubicle.17 However, it can be argued that observation of bedside staff resource within a fully operational critical care unit would be more determined by compliance with recommended standards and not a reliable measure of how much staff resource is actually needed. Observation of a ward area with lower nurse staffing levels than a critical care unit, and in which no prescribed nurse:patient ratio is enforced, is almost certainly a more valid environment in which to test whether being nursed in a cubicle consumes greater medical and nursing staff resource.
Currently, UK funding for HDC delivered outside a paediatric intensive care unit is generally absorbed into the ‘core’ main diagnosis HRG tariff, though there are a small number of stand-alone HDUs which receive additional funding for HDC HRG activity. This study supports a change to separate and specific funding for patients who meet HDC criteria, in line with suggestions of the importance of continually refining systems for activity based payments.27 It should also be considered in the context of recent work suggesting that specialist children's hospitals merit a surcharge of 20% in their HRG payments.28 Assuming a change in HDC commissioning would need to be cost neutral, provision of an additional tariff for all HDC activity would require a drop in core HRG tariffs to avoid double counting, but would enable payment to truly follow individual patients, reflecting differences in care provision.
Unanswered questions and future research
Further research could address the limitations of this work, including larger sample sizes with more individuals receiving rarer interventions, broader geographical spread of study sites and a more detailed ‘time and motion’ monitoring of staff time to include direct patient time not at the bedside. The question, however, is whether the considerable increase in resource intensity that would be required for such research, would lead to very different results and implications for policy, and thus whether the costs would be worthwhile. It may be that this work is sufficient to take forward the delivery of an additional HDU tariff for paediatric care in the UK and, indeed, in other settings, notwithstanding international differences in critical care provision.4
The study was funded through a grant from Birmingham Children's Hospital Research Foundation.
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Files in this Data Supplement:
- Data supplement 1 - Online supplement
Collaborators West Midlands Paediatric High Dependency Study Group. Steering group: Nicola Holdback (study co-ordinator), Kevin Morris (study CI), John Alexander, Joanna Coast, Raymond Oppong, David Whitehurst, Yvonne Heward, Chris Timmins, Phil Wilson, Heather Duncan, Fiona Reynolds, Ian Wacogne. Local study co-ordinators (medical and nursing): Birmingham Children's Hospital NHS Foundation Trust (Indra Van Mourik, Helen Cox; David Hobin, Mary Meehlhause; Phil Darbyshire, Jacky Hotchin; Ian Wacogne, Emma Dodson, Michelle Plaza; David Milford, Lorraine O'Reilly; David Barron, Rami Dhillon, Tracey Dawson, Jackie Clinton; Andrew Kay, Alia Murtaza; Tony Lander, Julie Cossey), Heart of England NHS Foundation Trust (Titus Ninan, Caroline Whyte), Sandwell and West Birmingham Hospitals (Chizo Agwu, Niten Makwana, Heather Bennett, Avnash Nanra, Joy Haywood, Paul Duflot, Rasekhuta Lephalala), The Royal Wolverhampton Hospitals (Penny Dison, David Coan, Karen Allen), University Hospital of Coventry and Warwickshire (Edward Simmonds, Leah Pritchard, Angela Sanderson), University Hospitals North Staffordshire (John Alexander, Caroline Whitehurst, Rose Burt), Worcestershire Royal Hospital (John Scanlon, Karen Haley). Observers: Cheryl Bates, Christopher Thorne, Sara Walker, Lea-ann Greaves, Jane Colley, Melanie Irving, Priya Deol, Reena Panchal, Helen Sinclair, Louise Perkins, Gemma Plant, Shelly Moore, Hayley Davies, Alex Purcell, Sarah Forrest.
Contributors KM, NH and JC conceived and designed the study. KM and NH co-ordinated delivery of the data collection. RO and JC undertook the analysis and health economic modelling. KM and JC drafted the initial manuscript. KM, RO, NH and JC revised the manuscript critically and approved the final version.
Funding The study was supported by a project grant from Birmingham Children's Hospital and a research grant from Birmingham Children's Hospital Research Foundation.
Competing interest None.
Ethics approval Hospital Trust R&D approval.
Provenance and peer review Not commissioned; externally peer reviewed.
Correction notice This article has been corrected since it was published Online First. One of the collaborators (David Whitehouse) has been corrected (to David Whitehurst).