Jump to: navigation, search

Acute Hospital Bed Days Per Capita

HQMNZ ID: HQM16.6.30.1449

Analytical Definition Document

Current version 0.4 25 May 2016

Document Version Management

Version Date Update Description Author

0.1 22/04/2016 First draft after 12 April 2016 meeting Alan Keegan

0.2 28/04/2016 Editing of the first draft Sylvia Yan

0.3 20/05/2016 Changes to reflect technical Group discussion Sylvia Yan

0.4 25/05/2016 Further feedback incorporated in particular 2.1.3.2, 3.1, 3.3.1.3 Sylvia Yan

Introduction: Acute Hospital Bed Days Per Capita

This document defines what acute hospital bed days per capita is and how it will be measured. Unless otherwise stated, this is referred to as the measure in this document.

Value and Rationale

Acute hospital bed days per capita is a measure of acute demand on secondary care that is amenable to good upstream primary care, acute admission prevention, good hospital care and discharge planning, integration of services and transitions between care sectors, good communication between primary and secondary care, can all help reduce unnecessary acute demand. Good access to primary and community care and diagnostics services is part of this. The value and the desired outcome of this measure are referenced through international practices with the following aims:


  • to understand demands on health system
  • to measure burden of care incurred by inpatients in the course of their hospital stay
  • to improve system-wide health services delivery
  • to ensure care appropriateness and efficiency
  • to support care in the home and community
  • to enhance partnership between hospital, primary and community care sectors.


The measure aligns well with the New Zealand Health Strategy's five themes, which are people-powered, closer to home, one team, smart system, and in particular value and high performance. They place an emphasis on measuring the performance of the whole system in order to determine the value the country receives from the system. The measure will be used to manage the demand for acute inpatient services on the health system. The intent of the measure is to reflect integration between community, primary, and secondary care and it supports the strategic goal of maximising the use of health resources for planned care rather than acute care. Different cohort groups of population such as Māori would have a high capacity to benefit from improved primary care that will reduce the need for acute care.

This measure is supported by a suite of locally selected contributory measures to strengthen the ability to detect and understand factors that drive acute demand. This combination of measures avoids the risk of a single high level measure which gives no indication of where improvements could be made. It also creates opportunities for inter-provider communication, and promotes data transparency and knowledge sharing.


High Level Defintion

The measure is the rate calculated by dividing acute hospital bed days by the number of people in the New Zealand (NZ) resident population. The acute hospital bed days per capita rates will be illustrated using the number of bed days for acute hospital stays per 1000 population domiciled within a District Health Board (DHB) with age standardisation.

This measure will be calculated quarterly with a rolling 12 month data period, as soon as the data is available. Results for quarter one of the 2016/17 financial year means the 12 months starting 1 July 2016 and ending 30 June 2017.

Acute hospital bed days are calculated by adding up the length of stay in days for patients presented to a NZ hospital acutely that are publicly funded. A stay will be counted in the measure if the first event in that stay is classified as an acute inpatient event.

This measure will be age-standardised and broken down by DHB of domicile. The data will be able to be stratified for other demographic variables.


Key Concepts

From the outset of this technical document three important concepts need to be understood. They are the concepts of Inpatient, Events (Common Counting Technical Advisory Group and Ministry of Health, 2014) and Stays (Ministry of Health, 2015). These concepts are consistent with the Acute Readmissions measure under development.

Inpatient

The broad concept of ‘Inpatient’ with the NZ public health setting refers to a patient whom has been admitted to a facility for the purposes of treatment and care for a period of time.

Events

An ‘Event’ refers to the period of care undertaken for a patient that has a specific start and end date and times. They identify the distinct elements of duration, period, facility, specialty and clinician responsible, diagnosis and treatments for the individual patient as well as many administrative attributes like funder, admission type etc.

Stays

The concept of a ‘Stay’ aligns largely with other analytical efforts with relevant contributory measures, such as Ownership OS3: Inpatient Average Length of Stay (ALOS) and Ownership OS8: Acute Readmissions under the DHB non-financial monitoring framework and performance measures framework. In this measure, it refers to a sequence of patient events (each with a distinct admit and discharge date and time) that can be logically joined together to form an inpatient ‘Stay’.


Two events are considered to be part of the same stay if the following criteria are met:

  • The first and second event have the same National Health Index (NHI),

and,

  • either the second event starts within 6 hours, or,
  • the first event has a valid transfer flag and the second event starts within 24 hours

Counting Acute Hospital Bed Days

Inclusions

The measure is meant to be comprehensive. So all types of acute stays should be included unless there is a strong rationale for exclusion. For the avoidance of confusion, the following are explicitly included:

  • maternity and neonatal services (tbc)
  • mental health services (tbc)
  • inpatient short stays at emergency department (ED) (tbc)
  • palliative care
  • scheduled cancer treatments R63/64
  • leave days (that is, leave day are not subtracted)
  • deceased patients
  • assessment, treatment and rehabilitation (ATR) beds, DSS214, HOP214, and HOP235 if it is part of a stay


The rule for these inclusions are in Section 2.1.

Exclusions

The following are explicitly excluded in the model:

  • Non-acute stays (the first event has a non-acute admission type)
  • Non-casemix stays (all events in the stay are non-casemix)
  • Overseas domiciled patients
  • Incomplete stays (last event in stay has a transfer flag)


The rule for these exclusions are in Section 2.2.

Hospital Data

The source of hospital data for the measure will be the National Minimum Dataset (NMDS). The events in NMDS will be used to count the numerator of this measure, acute hospital bed days. Events will be joined together into stays. The rules that define the joining together of individual patient events to form a patient stay are these:

  • The events are associated with the same person and have the same patient identifier i.e. NHI.
  • The events may be at facilities overseen by different DHBs of Service.
  • The prior event ends in a transfer i.e. each event in the sequence ends with a transfer indicator, aside from the last event.
  • There is less than 24 hours between the end of one event and the start of the next event
  • Sequential events that are up to 6 hours in the difference between the patient’s discharge time (of the prior event) and admit date and time, do not require a transfer flag to be deemed part of the stay.
  • Day Case events, wherever they occur in the sequence of events, are included
  • Non-casemix events that occur in the sequence are included.


All events within a stay will be counted as an acute stay if the first event is acute, regardless of whether any other events in that stay are acute or not. The following demographic variables will be included in the measure data:

  • DHB of domicile
  • age in 5-year bands (00-04, up to 84, and thereafter 85+)
  • gender
  • prioritised ethnicity (Māori, Pacific, and Other in that order)
  • NZDep2013 Index of Deprivation quintile

plus the following external field from the Primary Health Organisation (PHO) registers linked in using NHIs, last date of the stay against the quarter of the PHO enrolment data the patient is enrolled in:

  • PHO

Inclusions

Maternity and Neonatal Services

Description:

Maternity and neonatal services can be identified by specialty codes, purchase unit codes, diagnosis related groups (DRGs) and international classification of diseases codes (ICD).

Explanation:

Maternity services are provided in a range of settings. Hospitals are one of the many places this service is offered. There is concern that including maternity in the measure may promote perverse incentives against valid choices in providing maternity care.

The nature of the distribution of the length of stays in maternity is dependent on what alternative birthing facilities are available outside hospitals. -

Maternity is a large service and may have an influence over other elements in the measure. In practice, some births happen unexpectedly and acute beds are used to manage that. Distinctions between arranged and acute births is important, for example in the situation where a home birth has complications.

On the other hand, Lead Maternity Carers (LMCs) working in the community have large influence over maternity admissions. The measure needs to take a whole of system perspective and to be future proof to allow for integration and improvements.


Recommendation:

Maternity and neonatal services are included in the model.

A flag has been introduced using the health specialty codes, beginning with ‘P’, series for secondary maternity and Neonatal services. Further analysis will be carried out to better define these services in the model.

Mental Health Services

Description:

Mental health inpatient services is identified using a mental health specialty code with ‘Y’ prefix, Y00 to Y99.

Explanation:

Mental health services are an integral part of the system. Questions have been raised about whether mental health acute services have a different cost profile and resource utilisation compared to medical and surgical bed days.

There is a suite of Key Performance Indicators (KPIs) measures for mental health services which the Ministry and the sector are actively monitoring. Along with other key mental health initiatives and investments, these KPIs have proven to be effective in promoting better care management in the community settings. They could be served as contributory measures to help understand the drivers for acute hospital bed days demand.

Recommendation:

Mental health services are included. Forensic mental health services should be excluded.

A flag has been introduced using the health specialty code, being with ‘Y’, for mental health inpatient services. Further analysis will be carried out to better define these services in the model.


Inpatient short stays at emergency department

Description:

Inpatient short stays at emergency department, aka short stay EDs, describes hospital activities that have an event shorter than two days discharged by an emergency department specialist identified using health specialty code M05.

DHBs have different models of care. Generally all patients in an emergency department who received treatment for more than three hours would be required to be admitted as an inpatient event.


Explanation:

Some debates around whether these ED events should be excluded because of inconsistent practice of the admission rules for these events across DHBs. Some questions were asked about whether ED bed days are the same as other acute bed days and on how much of ED services are amenable to intervention.

ED is a hospital resource. The measure is about helping the sector to use health resources efficiently but the model should not disincentives models of care. One of the options is to perhaps moving short stay patient numbers to contributory measures. On the other hand, there are concerns about not counting this activity, than over weighting it.

Counting the length of stay of an event less than two days stay has been difficult prior to the introduction of the times stamps data (i.e. event start datatime and event end datetime) in NMDS in June 2011. Day cases or short stay EDs events had been excluded in analysis such as the historical Average Length of Stay measure. A constant such as 0.5 day had also been used in past analysis to accommodate these events.

Time stamps data has since improved and it is deemed to be suitable to be used to get fractional number of stays. Other alternatives and proposals could include having a minimum stay of half or one day. Excluding ED short stays may systematically exclude higher users of this service from the measure, e.g. Maori, and therefore may omit certain disparity issues to be addressed.

DHBs are encouraged to make improvements based on their own baseline. It is therefore less of a concern for inconsistency in reporting across DHBs. In practice, it is important to acknowledge that the measure cannot address everything and short ED stays may be a weakness of the measure.


Recommendation:

Short stay EDs are included. If Short stay EDs is in any part of the stay, then the stay is considered as a Short stay ED stay.

A flag has been introduced using the health specialty code M05 with less than two days length of stay, for Short stay ED services. Further analysis will be carried out to better define these services in the model.


Palliative Care

Description: Inpatient events that are discharged with the health specialty codes of ‘M80’ or ‘M81’ are considered as a palliative care event.


Two days is calculated based on the ‘length of stay’ field in NMDS and not the time stamps data

Explanation: Patients with terminal illness, undergoing end of life care can be admitted acutely as inpatients or non-acutely as part of the whole stay. The characteristics of the patients are significantly different from others in terms of medical nature and service required. However, these patients would benefit from having good non-hospital alternative care.

It is also technical difficult to identify Palliative care events accurate due to inconsistent recording of the service.

Recommendation:

Palliative care events are included in the measure.


Scheduled Cancer Treatments

Description:

Inpatients events with a DRG code starting with ‘R’, including ‘R63’ (Chemotherapy) or ‘R64’ (Radiotherapy).

Explanation:

Cancer patients can be admitted acutely for chemotherapy, radiotherapy and other procedures. The admission may be a scheduled treatment. The characteristic of this type of patients are likely to be different from other inpatients. In terms of clinical and health resource management, these patients cannot be simply excluded from admission analysis. These patients require adequate care planning both in and out of the hospital settings.

Recommendation:

Cancer treatments are included in the measure with the exception of same day cancer chemotherapy or same day radiotherapy events if these are a standalone event.


Event Leave Days

Description:

Event leave days are the number of days an inpatient is absent from the hospital at midnight.

Explanation:

Each DHB has their own models of care, so they have different policies on the use of event leave days. Excluding leave days would not accommodate DHBs that choose certain models of care for their patients.

Recommendation:

Leave days are included in the length of stay. That is, leave days should not be subtracted.


Deceased Patients

Description: Patients discharged as deceased. They are defined by the Event End Type Code of “DD”, “DO” or “ED”.

Explanation:

When a patient died in hospital, the patient was discharged as deceased with the Event End Type Code. Their stay in the hospital is a valid count for the measure. The patient however can then be acutely readmitted to donate organs for transplantation purposes. The second event can be mistakenly counted as an acute admission. The number of events with this scenario is insignificant and will not have much influence over the measure calculation.

Recommendation:

Patients discharged as deceased are included in the measure.


Exclusions

No-acute stays

Description:

A non-acute inpatient stay is excluded when the first event of the stay is not an acute admission. For example admission type of the first event is coded as ‘AA’ or ‘WN’, rather than ‘AC’.

Explanation:

A key element of an acute readmission stay is that the first event in the readmission sequence has to be an unexpected emergency inpatient event. This exclusion removes patient stays that the first events are admitted as elective or arranged.

Recommendation: Non-acute stays are not included in the numerator of the measure.


Non-casemix stays

Description:

A non-casemix inpatient stay only involves a single or multiple inpatient events that all are discharged as non-casemix events. It is defined by the purchase unit (PU) code of “EXCLU”. None of the events that formed the stay was discharged with a casemix PU code.

ATR events, but not the entire stay, with PU codes of ‘DSS214’,’HOP214’ and ‘HOP235’ or mental health non-casemix services are included if they are part of a casemix stay.

Explanation:

Inpatient events and stays need proper comparison between facilities for the variety of patient activity and reporting methods across DHBs. While the current casemix approach is primarily used for funding purposes, this approach can also be applied on inpatient stays to ensure that inpatient stays are relatively homogenous across the sector.

Recommendation:

Non-casemix stays, not the individual non-casemix events as part of a casemix stay, are not included in the numerator of the measure.


Overseas patients

Description:

Patients who usually live overseas, as defined by their residential address and identified by patient domicile code of ‘999’ are excluded.

Explanation:

The measure’s intent is to manage demand for acute care. The main way DHBs should manage acute care demand is improvements to quality and coverage of primary care. DHBs have constraints managing primary care for overseas residents. So these patients should not be included in the measure.

Recommendation:

Overseas patients are not included in neither the numerator, nor the denominator, of the measure.


Population Data

Population Denominator

Population projections of DHB estimated resident population (ERP) supplied by Statistics New Zealand will be used as the denominator of age-specific rates. This is consistent with the population used in the Population Based Funding Formula (PBFF). The ERP was chosen over alternatives (e.g. PHO patient registers) because it reflects DHBs and PHOs have responsibility to everyone, whether they are enrolled in a PHO or not. The PBFF population series provides the best estimate of NZ and region population numbers, and it is suitable to be used to reflect the impact of Regional Alliances in their areas.

PHO Population

This measure is intended to be used to promote integration between primary care and hospital care. Although the PHO patient registers are not used as the denominator for the measure, it is still important to demonstrate its linkages by identifying which PHO the patients are enrolled in. So that local initiatives could be targeted to specific cohort group of patients.

NHIs will be used to link NMDS data with the PHO patient registers. The PHO patient registers for the quarter in which the patient was discharged will be used. For example, if the patient’s last day of the whole stay was on 2 February 2016, then the PHO the patient is enrolled with in that quarter will be used. In other words, the PHO enrolment data for the quarter ending March 2016 will be used to attribute the total bed days of that stay to.

Standardisation

Description:

Standardisation is often used when comparing rates between different populations, for example different countries or DHBs. Rates are often standardised by age because the prevalence of many conditions is associated with age which makes age one of the major confounding factors for explaining differences between groups. For example, older people are more likely to get long term conditions than younger people. Using age standardisation allows different populations (for example different countries or DHBs) to be compared on a more comparable basis by reweighting their age-specific rates to a reference population.

Explanation:

The draft model includes both crude and age standardised rates. The rationale for not standardising by other demographic factors such as ethnicity, gender or deprivation was that it may compromise the addressing of health inequity. However, the data will be able to be stratified for these factors to enable analysis of their impact and inform interventions.

The WHO standard population (World Health Organization, 2001) along with the Statistics New Zealand projected population are commonly used in New Zealand as a reference population for age standardisation. New Zealand Māori Population has been used for initiatives addressing specifically indigenous inequality.

While the choice of reference population is mostly subjected to interest, there is some concern that a population like WHO will systematically mask Māori health inequalities. This is because the Māori population has a much younger age structure than globally oriented populations. However, others have raised concerns that WHO population introduce an artificial weighting towards younger population which generate unfair rates for older population.

At a principle level, a system level measure needs to take into account all different types of disparity such as ethnicity, age, gender, social economic factors in order to reflect a whole of system performance. Given that DHB is the key comparative unit, a national population that reflects the average age structure of majority of DHBs is deem to be most suitable.

Recommendation:

The data will be age-standardised only and the estimated NZ resident population with Statistics NZ projections has been used in the model for direct standardisation. The data will be able to be stratified for different demographic groups.

Age standardised rates using WHO standard population and NZ Maori Population have also been included to provide comparison. Feedback from the sector on this will inform further discussions.


References

Common Counting Technical Advisory Group and Ministry of Health. (2014, January). Common Counting Standards 2013–14. Retrieved April 22, 2016, from Nationwide Service Framework Library: http://nsfl.health.govt.nz/purchase-units/common-counting-standards-2013%E2%80%9314

Ministry of Health. (2015, December). Performance measures for 2015/16. Retrieved April 22, 2016, from Nationwide Service Framework Library: http://nsfl.health.govt.nz/accountability/performance-and-monitoring/performance-measures/performance-measures-201516

Technical Working Group - Acute Readmissions. (2016, April 22). Technical Working Group - Acute Readmissions - Analytical Definition Document. Wellington, New Zealand: Ministry of Health.

World Health Organization. (2001). Age standization of rates: A new WHO standard. Retrieved April 21, 2016, from World Health Organization: http://www.who.int/healthinfo/paper31.pdf