Stats NZ has a new website.

For new releases go to

www.stats.govt.nz

As we transition to our new site, you'll still find some Stats NZ information here on this archive site.

  • Share this page to Facebook
  • Share this page to Twitter
  • Share this page to Google+
8 – Data availability: health care

8.1 Health care: output quantity

This section sets out the data sources that exist for health care services and accompanying analyses in New Zealand, and discusses the pros and cons of each of the sources from the perspective of output measurement.

A great deal of information is collected every day on the health care services provided in New Zealand, whether in primary or secondary care. Much of this information covers the whole of the national territory and is made available centrally. These sources are known collectively as the ‘National Collections’ and are:

  • General Medical Subsidy Collection (GMS)
  • Health workforce information
  • Hepatitis B Screening programme (Hep B)
  • Laboratory Claims Collection (Labs)
  • Maternity and Newborn Collection (MNIS)
  • Medical Warnings System (MWS)
  • Mental Health Information Collection (MHDW)
  • Mental Health Information National Collection (MHINC)
  • Mortality Collection
  • National Booking Reporting System (NBRS)
  • National Booking Reorting System Data Warehouse (NBRS DW)
  • National Health Index (NHI)
  • National Immunisation Collection (NIR)
  • National Minimum Dataset (Hospital Events) (NMDS)
  • National Non-admitted Patient Collection (NNPAC)
  • New Zealand Cancer Registry (NZCR)
  • Pharmaceutical Collection (Pharms)
  • Primary Health Organisation Enrolment Collection (PHO)

Appendix 3 repeats this list, providing brief descriptions for all in the interests of comprehensiveness. These National Collections have not been designed for measuring productivity, but nevertheless many of them include relevant information. Those collections that are listed above in bold are those which offer useful information for productivity measurement, and these are the sources that are described more fully below.

The Mortality Collection and the National Health Index are sources that would prove useful if quality adjustment (based on changes in mortality rates) and a health care pathway approach (using the NHI to link records) were to be worked on.

Other information is collected outside of the framework of the National Collections. These are covered in section 7.1.9.

A word of caution is needed on the occasional duplication of an activity in two or more of these distinct national collections. The reasons for the duplication in separate databases is due to the particular use of each database. For example, hospital discharges associated with newborns and maternity services appear in both NMDS (in order to record all hospital activity) and the Maternity Newborn Collection (MNIS) (in order to be comprehensive about newborn information). The appearance of the National Health Index (NHI, this is an identifier unique to any individual) on the separate records, along with event date, is the key to avoid double-counting.

There are also analyses of these raw data which are used, or could be used, in monitoring output, inputs, or productivity.

top

8.1.1 National Minimum DataSet (NMDS)

Overview

The NMDS is a national collection of public and private hospital discharge information, including clinical information, for inpatients and day patients (day care). Data have been submitted electronically by public hospitals since 1993 and for publicly funded events in private hospitals since 1997. Extracts available to MoH include calculated variables, such as cost weight and DRG. Customised data analysis, summarising and reporting can be requested from the Ministry of Health’s Information Directorate, and analytical event files are regularly downloaded including those used to calculate the inpatient output component of the Ministry’s productivity metric.

Coverage

NMDS captures data on all patients discharged from day patient and inpatient care in publicly funded hospitals and publicly funded patients in private hospitals throughout New Zealand. It does not cover privately funded activity in private hospitals.

An entry in NMDS corresponds with a single discharge. Re-admissions, transfers (to another hospital) etc can be identified through the use of the National Health Index (NHI) which is coded to all records. A set of assumption-based rules may be required to ‘link’ separate discharges into a single treatment. For example, a time limit may be needed to distinguish between (i) re-admission for the same bout of illness and (ii) repeat occurrence of the same health problem (after complete recovery of the initial bout).

Timing (frequency, time reference, length of time series)

NMDS is a continuous 100 per cent survey of all inpatient and day patient discharges.

MoH holds NMDS datasets that include casemix weights (from the WIES system) for 6-monthly periods, either January to June or July to December.

Data on discharges from public hospitals as far back as 1988 are stored in the NMDS, although it is noted that there have been ‘many changes over the years’, including addition and deletion of variables as well as in scope/coverage. Data on publicly funded discharges from private hospitals has been collected since 1997.

Use in a measure of productivity

This source provides information on the number of inpatient and day patient activities in public and private hospitals, by type of activity (for example by ICD and DRG) along with information on estimated cost, according to the WIES costing system. For this reason, the NMDS has been, and continues to be, the main data source for existing productivity estimates published by the Ministry of Health. See section 4.3.1 for further information on the Ministry of Health’s productivity estimates.

Account needs to be taken of a number of factors which will affect whether, and how, each record is processed from the perspective of use in an output metric. The key consideration will be how the record corresponds to a unit of output:

‘1 record = 1 unit of output’: in many cases, the record will describe a single activity which constitutes the whole course of treatment within the hospital environment for a patient with a particular diagnosis. In such cases, the record may correctly be considered a unit of output in its own right.

‘Concatenation of records’: the availability of the NHI on all records means that multiple activities for the same diagnosis can (should) be linked together, allowing a health care pathway type approach to be compiled. Clearly, one record in such cases does not correspond with one unit of output, but the single unit of output will be made up of a number of separate records.

‘Exclusion of records’: some records may not correspond to ‘output’ at all and should not be included in a measure of output. Examples of this may be cancelled appointments or patients who die whilst admitted. Care needs to be taken in determining whether there is indeed ‘output’ generated in such cases, as in some cases the risk of death is high and the hospital is trying its best to prolong life under difficult circumstances; this type of output might be likened to the services of a barrister when the defendant is found guilty.

Known issues

None.

Key variables top

The following variables (in bold), along with short descriptions, are those which should be considered useful when analysing information in the NDMS for measuring change in output.

  • Admission source code This variable helps identify activities along the health care pathway (it identifies transfers from other hospitals).
  • Age at admission and age at discharge and CCL (complication/co-morbidity class level) These may be of use if it is considered desirable to distinguish between patient of different types (for example, older patients tend to require more resources than younger patients through earlier admission etc).
  • Costweight This is the relative price (see section 7.3.1 for a discussion of use of this in measures of output change).
  • Diagnosis and clinical code (in terms of the International Classification of Disease ICD) These variables identify the primary and other diagnoses for the admitted patient.
  • DRG code current This is the key for identifying multiple activities within a single patient’s health care pathway.
  • Encrypted NHI number This is the key for identifying multiple activities within a single patient’s health care pathway.
  • Event end date and Event start date These variables help to identify the sequence of activities whining a health care pathway
  • Event end type code and death flag This variable helps to identify activities which should be linked as part of the health care pathway. It would identify activities resulting in death, if it is considered that such activities should not form part of ‘output’.
  • Health specialty code This variable may form part of the disaggregation classification and help to distinguish between different types of health activity.
  • Length of stay This may help in quantifying output, for types of care for which the unit of output is considered to be ‘a week’s worth of care received’. It may also be of use in distinguishing between different types of care (if ‘long’ lengths of stay are considered to constitute a different type of output from ‘short’ lengths of stay).
  • NZ resident status This variable helps to identify activities that are within scope (for example in order to exclude treatment for non-residents I required).
  • Principal health service purchaser This helps identify the scope of the output measure, by distinguishing between, for example, ACC funding, private funding etc.
  • Private flag This variable will help identify activities which are within scope (for example indentifying patients paying privately).

Level of disaggregation

NMDS includes a number of variables which capture information on the characteristics which might be considered to be important to the consumer. These include:

Age at admission, age at discharge, CCL (complication/co-morbidity class level), DRG code current, event end type code, health specialty code, and length of stay.

The disease classification at its lowest level is too fine, with over 10,000 categories, which would lead to some cells having zero activity recorded for some years. A problem with some of the other classifications, such as DRG, might be associated with the extent to which homogenous activities are spread across different DRG categories, whereas the ideal situation would involve grouping such activities. For example, the treatment of mental health issues may take the form of counselling or prescription of pharmaceuticals, or a combination of both. See section 3.1.3 ‘Level of disaggregation for the measure of output’ for a discussion of the level of disaggregation.

Overlaps / duplication with other sources

For some kinds of discharge, the record of the discharge will, or may, appear in other databases. The appearance of the NHI on the separate records is the key to avoid double-counting. The reasons for the duplication in separate databases is due to the particular use to which each database is put. For example, hospital discharges associated with newborns and maternity services appear in both NMDS (in order to record all hospital activity) and the Maternity Newborn Collection (MNIS) (in order to be comprehensive about newborn information).

Access to dataset

The MoH’s 6-monthly datasets are held as SAS files, and are relatively large: each file is some 5 MB on average. Processing and analysis within MoH is carried out within the SAS environment (currently SAS 9.1).

The patient identifier, the NHI, is encrypted.

Corresponding weights

Estimated costs are calculated by the WIES (Weighted Inlier Equivalent Separations) system. The methodology and data sources are complicated and complex and accordingly are covered separately in section 7.3.1.

top

8.1.2 National Non admitted Patient Collection (NNPAC)

Overview

NNPAC stores data about non admitted face-to-face secondary care events, such as outpatient and emergency department visits. The database records both first and follow-up appointments. The main purposes are to monitor non-admitted patient events, to analyse inter-district flows and to monitor the impact of policy. Unlike NMDS, NNPAC does not include information on associated costs, nor on diagnoses or procedures which could be used for disaggregation.

Coverage

NNPAC is designed to be a comprehensive store of all non-admitted events in public hospitals. It also includes information on non-attendances (where the appointment was not cancelled but the patient either never arrived or left before seeing the doctor).

Timing (frequency, time reference, length of time series)

NNPAC is only a recently established database, being established in 2006 and contains data from July 2005. It is a continuous 100 per cent survey.

Use in a measure of productivity

NNPAC provides information which covers other secondary care activity beyond what is already recorded in NMDS, although care needs to be taken as there are records which of the same activity appearing in both NMDS and NNPAC. The NHI is the key to dealing with this double counting.

Known issues

MoH reports that there are some issues about comprehensiveness, due to a fair degree of non-compliance. This could be a significant problem, as it makes measuring volume change over time rather more difficult. For example, if there was a 10 per cent increase in the number of appointments from one year to the next, how much of this would be due to general practitioners seeing more patients and how much would be due to general practitioners recording more of the patients that they see?

It would be fairly easy to deal with the activity of general practitioners who have not recorded appointments before and start all of a sudden: these would be ignored for the first year, and introduced into the calculations as of the second year in order to create a matched pairs comparison. Care would need to be taken with ongoing improvement or deterioration in recording, whereby it may be difficult to distinguish between increasing (decreasing) quality of reporting and increasing (decreasing) activity.

The information held by NNPAC on renal dialysis services is thought to be problematic, with a single unit of measurement not consistently used (one ‘unit’ may refer to a single treatment, or it may be to a set of treatments).

For renal dialysis and oncology services, there is occasional duplication of records between NNPAC and NMDS, which can on the whole be identified using the NHI.

Key variablestop

The following variables (in bold), along with short descriptions, are those which should be considered useful when analysing information in NNPAC for measuring change in output.

  • Age at time of visit separating activity according to age (or at least bands of age such as ‘young’, ‘adult’, ‘old’) may pick up some of the characteristics of patients when they present (older patients typically require more health care resources for the same diagnosis).
  • Attendance code This variable identify whether a patient attended or not, and therefore helps identify whether the record constitutes a unit of output or not.
  • Encrypted NCU id (NHI) This is the key for identifying multiple activities within a single patient’s health care pathway.
  • Equivalent purchase unit This variable identifies which type of contract the event is funded under, and helps associate a price to the event.
  • Event type This variable may be of use in distinguishing between different types of activity.
  • Health provider type This variable distinguishes between the type of staff providing service: doctor, nurse or other, and may be of use in distinguishing between different types of activity.
  • Health specialty code This variable may form part of the disaggregation classification and help to distinguish between different types of health activity.
  • Purchaser code This variable identifies who has paid for the event, and will help to screen out, for example, overseas patients paying privately, if required.
  • Service type This variable distinguishes between first and follow-up appointments, and may be of use in moving away from an activity-based unit of output to a health care pathway based unit of output. Combining records for first and (multiple) follow-up appointments together into a single unit of output would approximate a health care pathway if the pathway only involved such appointments. This will not be the case if the patient’s health care pathway also includes other activities, for example as an inpatient attendee or if the patient sees a general practitioner.
  • Volume This variable will help to identify the relative weight for an event. In general, the volume is recorded as ‘1’, but may be a fraction in cases where the unit of measurement for the type of activity in NNPAC is, say, a block contract purchase for a number of separate units of treament.
  • NZ resident status This variable helps to identify activities that are within scope.
  • Private flag This variable will help identify activities which are within scope.

Level of disaggregation

NNPAC includes a number of variables which capture information on the characteristics which might be considered to be important to the consumer. These include: age at time of visit, attendance code, equivalent purchase unit, event type, health provider type, health specialty code, and service type.

Access to dataset

MoH has access to the dataset for analytical purposes.

Corresponding weights

NNPAC does not include information on relative weights for the events included.

For some types of outpatient and emergency department activity, WIES are available, and may provide sufficient information (see section 7.3.1).

Outpatient and emergency department activity is funded through contracts, called District Annual Plans (DAPs) between DHBs and the Ministry. Each DHB negotiates its own DAP with the Ministry on an annual basis, and the number of each different type of activity, along with the price, is set out in part of the DAP called the Price Volume Schedule (PVS). In general, there is little notion of a nationwide price for activities, as each DHB negotiates individually. As the PVS is part of the DAP, the number of activities and prices are on a planned basis: there is no reporting of the corresponding turnout at a disaggregated level (planned activity / expenditure and actual turnout is reconciled at the level of total expenditure).

The PVS includes a set of ‘adjusters’ which are designed to modify the turnout prices in the case of less or more activity being carried out.

 Further Analysis needed 1

Where both exist, the WIES and the average DAPs / PVS prices for NNPAC activity should be compared with the aim of choosing the most suitable weights for outpatient and emergency department activity.

top

8.1.3 General Medical Subsidy Collection (GMS)

Overview

GMS contains information on the fee-for-service payments made to doctors for patient visits that have been processed by the HealthPAC Proclaim system; that is, visits to doctors other than the one with which the patient is enrolled. GMS is used to monitor contracts with providers, support forecasting and setting of annual budgets, and analyse health needs and assess policy effectiveness.

Coverage

GMS includes information on fee-for-service payments to doctors other than the one with which the patient is enrolled. It also includes after-hours visits for both enrolled and non-enrolled patients. Most GP visits for which there are no fee claims are excluded from GMS.

Timing (frequency, time reference, length of time series)

GMS was established in August 2003 and contains data from November 2001.

Use in a measure of productivity

GMS includes information that could be used as a count of the number of publicly-funded (publicly-subsidised) GP appointments.

The inclusion of the NHI (the unique patient identifier in NZ) would allow appointments for the same health care treatment to be aggregated up from single, independent ‘activities’ to health care pathways. For patients with co-morbidities in particular (but not only), this variable would need to be used in conjunction with other variables, such as date of event and diagnosis, in order to ensure that the correct set of activities are aggregated in the appropriate health care pathways.

Known issues

The MoH reports that there are some issues about comprehensiveness, due to a fair degree of non-compliance (non-submission of information by what’s thought to be 40 percent of Primary Health Care Organisations). This could be a significant problem, as it makes measuring volume change over time rather more difficult. For example, if there was a 10 percent increase in the number of appointments from one year to the next, how much of this would be due to general practitioners seeing more patients, and how much would be due to general practitioners recording more of the patients that they see?

One method of dealing with the activity of general practitioners who have not previously recorded any appointments and then start to record all appointments, would be to ignore activity for the first year, and begin comparison from the second year onwards in order to create a matched pairs comparison (and vice versa). Care would need to be taken with ongoing improvement or deterioration in recording, whereby it may be difficult to distinguish between increasing (decreasing) quality of reporting and increasing (decreasing) activity.

Key variablestop

The following variables (in bold), along with short descriptions, are those which should be considered useful when analysing information in GMS for measuring change in output.

  • Age in years Separating activity according to age (or at least to bands of age such as ‘young’, ‘adult’, ‘old’) may pick up some of the characteristics of patients when they present (older patients typically require more health care resources for the same diagnosis).
  • Amount paid (excluding GST) This variable could be used as part of a set of information used in constructing the relative weight. Information on the patient co-payment (and other contributions to total cost) would also be needed to ensure that the relative weight is correctly specified.
  • Encrypted NHI number This is the key for identifying multiple activities within a single patient’s health care pathway.
  • Health professional group code This variable distinguishes between the professional accreditation of staff providing service: Medical Council, Nursing Council, and Dental Council, and may be of use in distinguishing between different types of activity.

Level of disaggregation

GMS includes a number of variables which capture information on the characteristics of the service which might be considered to be important to the consumer. These include: age in years and health professional group code.

Access to dataset

The MoH has access to the dataset for analytical purposes.

Corresponding weights

GMS includes information on the payment made to the medical practitioner as part of the fee-for-service. In order to use this information in constructing a relative weight for these kinds of activity, the total amount of payment needs to be known. This can be compiled using information on the capitation payment in the PHO collection (see section as well as the amount of the patient co-payment.

top

8.1.4 Primary Health Organisation collection (PHO)

Overview

PHO contains information on patients enrolled with general practitioners, along with information on the capitation payments paid by the Ministry as part of the overall funding system for general practitioners and PHOs. It is used to assist PHOs, DHBs, and the Ministry to report and monitor patient enrolment, to provide PHOs, DHBs, the Ministry, and researchers with population data to assist with population health research, and to assist PHOs to examine and improve the quality of their enrolment information.

Coverage

PHO captures information on all patients who have enrolled with a general practitioner.

Timing (frequency, time reference, length of time series)

PHO was established in 2005 and is updated quarterly.

Use in a measure of productivity

PHO does not contain explicit information on activity, although it does include the date of the last contact with a primary health care doctor in that quarter, which can be used to contribute to a count of contacts in conjunction with other sources (GMS, Labs, and Pharms: see their entries in this section).

PHO also includes information on whether or not enrolled patients are also enrolled as Careplus patients (see Careplus enrolment status variable below for definition), which could be one component used in constructing health care pathways.

A third use in measuring output quantity, is that some of the payment made to PHOs reflects achievement against public health targets: the nature of these targets, along with the potential for compiling indicators of associated output, has yet to be explored. The PHO Performance Management weblink below contains a document explaining the performance indicators and payments available to PHOs.

http://www.dhbnz.org.nz/Site/SIG/pho/default.aspx

PHO also contains information on capitation costs, which should be used in constructing the weights for GP activity. Capitation payments are dependent on a number of variables, including age of patient, whether or not certain types of service card (for example, a community service card) are held, deprivation of the area and ethnicity of the patient, as well as whether or not some payment has already been made through the GMS system. These payments should be added to patient and DHB fee-for-service co-payments for GP appointments to arrive at total cost.

Known issues

None.

Key variablestop

The following variables (in bold), along with short descriptions, are those which should be considered useful when analysing information in PHO for measuring change in output.

  • Age Separating activity according to age (or at least to bands of age such as ‘young’, ‘adult’, ‘old’) may pick up some of the characteristics of patients when they present (older patients typically require more health care resources for the same diagnosis).
  • Amount payable, actual ffs deduction amount, and fee amount These variables could be used as part of a set of information used in constructing the relative weight for GP appointments. The actual amount paid from the capitation funding side is the difference between what is due because of the capitation formula for the particular PHO and what has been paid through the GMS system for patients enrolled with that same PHO, but who had an appointment with a doctor other than the one the patient is enrolled with.
  • CBF NHI number This is the key for identifying multiple activities within a single patient’s health care pathway.
  • Careplus enrolment status This variable identifies whether a patient is a Careplus enrollee or not. Careplus is a funding programme for PHOs designed to provide low cost access for people with high needs in New Zealand. This variable may help distinguish between single activities and health care pathways.

Level of disaggregation for a measure of output quantity change

PHO includes two variables which capture information on the characteristics of the service which might be considered to be important to the consumer. These include: age and Careplus enrolment status.

Access to dataset

The MoH has access to the dataset for analytical purposes.

Corresponding weights

See Amount payable, actual ffs deduction amount, and fee amount in the Key variables section above.

top

8.1.5 Primary Health Organisation high-level volume reporting

Overview

Some PHOs provide the Ministry with a high-level report setting out the volume of primary care contacts each quarter. Data are broken down by age, ethnicity, and deprivation.

Coverage

65 of the 91 PHOs provide this information.

Timing (frequency, time reference, length of time series)

PHOs provide this high-level information once a quarter, with the data relating to total contacts in the quarter.

Use in a measure of productivity

This may be a good source with which to corroborate the estimates of primary care activity compiled through other sources. If deemed sufficiently good, it could even be the preferred source of information on change over time, given the patchy nature of other sources of information.

Known issues

Only a subset of PHOs provide the information, so there may be some selection bias inherent in any estimates of change in activity if, for example, the subset of PHOs are not representative.

Only counts of contacts are available, so there is no possibility of aggregating up to health care pathways, or formulating any kind of unit of output other than activity.

Access to dataset

The MoH has access to the dataset for analytical purposes.

Corresponding weights

Given the lack of disaggregation in the information reported, weights would simply be total GP appointment weights, taken from the same source(s) as for PHO (see section 8.1.4).

top

8.1.6 Laboratory claims collection (Labs)

Overview

The Labs collection holds information on publicly funded primary care tests. Until 2008, Labs was the payment system for ensuring funds flowed to the correct test provider. This is no longer the case, and Labs is now only a data warehouse without the direct role in payment processing. For tests carried out prior to 2008, Labs contains claim and payment information for primary care test subsidies that have been audited against the HealthPAC Proclaim system (as well as those reported directly by a subset of DHBs). For tests carried out after 2008, Labs continues to be the data warehouse for information on the primary care tests paid for using public funds., although it has been suggested that the data warehouse is no longer comprehensive due to falling data provision rates.

There are known quality issues with the way payments reported prior to 2003 were allocated to individual DHBs, although this should have no impact at the national level.

Labs allows the Ministry of Health and DHBs to monitor primary care test subsidies.

Coverage

Labs includes information on primary care test subsidies reported in HealthPAC and by a subset of DHBs. It is not comprehensive, so changes over time need to be configured carefully to distinguish between changes in coverage and change in volumes of matched pairs.

Timing (frequency, time reference, length of time series)

Labs was established in 2000 and contains data from July 1997.

Use in a measure of productivity

Labs includes information that could be used as a count of the number of publicly-funded (publicly-subsidised) laboratory tests.

The inclusion of the NHI (the unique patient identifier in NZ) would allow lab tests to be aggregated up with other components of a health care pathway. For patients with co-morbidities in particular (but not only), this variable would need to be used in conjunction with other variables, such as date of event and diagnosis, in order to make sure that the correct set of activities are aggregated in the appropriate health care pathways.

Labs also includes the date of referral, which can be used as part of the count of general practice activity – where this date is different from dates already recorded in other sources such as GMS, PHO, and Pharms (see sections 8.1.3, 8.1.4, and 8.1.8).

Known issues

The recent change in the status of Labs, from a key component of the payment system to data warehouse, has meant that there has been a reduction in the comprehensiveness of the recording of tests.

Key variablestop

The following variables (in bold), along with short descriptions, are those which should be considered useful when analysing information in Labs for measuring change in output.

  • Age at visit Separating activity according to age (or at least to bands of age such as ‘young’, ‘adult’, ‘old’) may pick up some of the characteristics of patients when they present (older patients typically require more health care resources for the same diagnosis).
  • Amount paid EXCL (excluding GST) This variable could be used as part of a set of information used in constructing the relative weight. Information on the patient co-payment (and other contributions to total cost) would also be needed to ensure that the relative weight is correctly specified.
  • Encrypted NHI number This is the key for identifying multiple activities within a single patient’s health care pathway.
  • Laboratory test, test code and laboratory test group These variables may form part of the disaggregation classification and help to distinguish between different types of health activity.
  • Number of tests This helps to correctly distinguish between price and volume: each time a test is carried out, the reimbursement is paid.
  • Provider type This variable distinguishes between the professional accreditation of staff providing service: Medical Council, Nursing Council, and Dental Council, and may be of use in distinguishing between different types of activity.
  • Referral ID and visit date These variables would help in linking the lab test with other components of the health care pathway, and in identifying the total number of GP contacts.

Level of disaggregation

Labs includes a number of variables which capture information on the characteristics of the service which might be considered to be important to the consumer. These include: age in years and health professional group code.

Access to dataset

The MoH has access to the dataset for analytical purposes.

Corresponding weights

Labs includes information on the payment made to the provider.

top

8.1.7 Mental Health Information National Collection (MHINC)

Note: as of 1 July 2008, this database migrated to PRIMHD.

Overview

MHINC draws together information on the provision of secondary mental health and alcohol and drug services purchased by the government. This includes secondary inpatient, outpatient, and community care provided by hospitals and non-government organisations (NGOs). MHINC does not include information on the provision of primary care mental health services, for example, by general practitioners.

Coverage

MHINC includes information on secondary inpatient, outpatient, and community care provided by hospitals and non-government organisations, but not primary care mental health services, for example, by general practitioners.

Due to some psychogeriatric services being funded differently than other mental health services, these are not captured in this database. Also, some mental health services that are funded under block or bulk contracts are not captured in MHINC.

Timing (frequency, time reference, length of time series)

MHINC was started in July 2000. Information on services provided by NGOs were not completely recorded in early years, in particular before 2003.

Use in a measure of productivity

MHINC includes some information that is not already covered in either NMDS or NNPAC, but care needs to be taken as these other databases will record activity where appropriate (for example, if the event is as a result of the patient being admitted, then NMDS will capture this).

Care will need to be taken to ensure that an appropriate unit of output is used. For example, a unit of output defined according to number of visits will be subject to measurement bias if clinical guidance on frequency of visit changes over time. A more appropriate unit of output for mental health may be ‘management of patient over a period of time’.

MHINC does not include any information on relative weights for the different events captured.

Known issues

As of 1 July 2008, this database migrated to PRIMHD.

Key variables top

The following variables (in bold), along with short descriptions, are those which should be considered useful when analysing information in MHINC for measuring change in output.

  • Service setting code, service code, admission type code These variables may be of use in distinguishing between different types of activity, and may be of use in screening out activity already captured in NMDS and NNPAC.
  • Units of service This variable defines what the unit of measurement in the database is; for example, number of bed days, number of attendances.
  • Agency code This variable identifies the type of provider of service, for example DHB provider arm or NGO, and may be of use in setting the scope of the output measure.
  • Date of birth Separating activity according to age (or at least to bands of age such as ‘young’, ‘adult’, ‘old’) may pick up some of the characteristics of patients when they present (older patients typically require more health care resources for the same diagnosis).
  • Diagnosis (principal and additional), health specialty code, and clinical code (in terms of the International Classification of Disease (ICD)) These variables may form part of the disaggregation classification and help to distinguish between different types of mental health activity.
  • Discharge type code, referral date, referral type code, event end type code, event end date These variables may be of use in moving away from an activity-based unit of output to a health care pathway based unit of output, as they identify, for example, end of treatment or transfer.
  • Encrypted NHI number This is the key for identifying multiple activities within a single patient’s health care pathway.
  • NZ resident status This variable helps to identify activities that are within scope.
  • Principal health service purchaser This variable identifies who has paid for the event, and will help to screen out, for example, overseas patients paying privately.

Level of disaggregation

MHINC includes a number of variables which capture information on the characteristics which might be considered to be important to the consumer. These include: service setting code, service code, admission type code, diagnosis (principal and additional), health specialty code, and clinical code

Overlaps / duplication with other sources

Where appropriate, mental health events are recorded in NMDS and NNPAC, and therefore care will need to be taken to ensure that events are not double counted when drawing together information on events recorded in the three databases.

Access to dataset

Via the MoH.

Corresponding weights

MHINC does not include information on relative weights for the events included. The National Pricing Project (NPP) identifies the costs of these events.

top

8.1.8 Pharmaceutical Collection (Pharms)

Overview

Pharms is a data warehouse that supports the management of pharmaceutical subsidies. It contains claim and payment information from pharmacists for subsidised dispensing that have been processed by the HealthPAC General Transaction Processing System (GTPS).

Coverage

Pharms contains claim and payment information from pharmacists for subsidised dispensing.

Timing (frequency, time reference, length of time series)

Pharms was started in July 1992, although records prior to 1996 have been archived (they are available on request).

Two major changes have been introduced since 1992:

  • Repeat prescriptions were introduced in 1996; and
  • The major drug key changed from Medicode to Pharmacode in 1998.

Use in a measure of productivity

Pharms provides information on pharmaceutical products that can either be seen as units of output in their own right, or components of a health care pathway (alongside GP appointments and so on).

Care will need to be taken to ensure that an appropriate unit of output is used. For example, as clinical guidance changes over time, prescribing practice may change such that in one year a 100mg dosage of a given pharmaceutical for a particular patient’s condition is prescribed 6 times a year, and the next year a 50mg dosage is prescribed 12 times a year. In both years, the same quantity of pharmaceutical is prescribed. Different definitions of the unit of output will result in different estimates of volume change.

The ability to distinguish repeat prescriptions allows analysis of different units of output. See text against the Repeat sequence number variable in the key variables section below.

Pharms also includes the date of dispensing, which can be used as part of the count of general practice activity (where this date is different from dates already recorded in other sources such as GMS, PHO, and Labs).

Known issues

None.

Key variables top

Pharms has a vast range of variables, many of which could be useful in productivity analysis. In short, the most important variables are:

  • Age at dispensing Separating activity according to age (or at least to bands of age such as ‘young’, ‘adult’, ‘old’) may pick up some of the characteristics of patients when they present (older patients typically require more health care resources for the same diagnosis).
  • Date dispensed This variable would be of use if aggregating the pharmaceutical costs with other costs along the health care pathway, and in identifying total number of GP contacts.
  • Encrypted NHI number This is the key for identifying multiple activities within a single patient’s health care pathway.
  • Formulation ID This variable identifies the chemical, and is probably the main way to distinguish between different types of pharmaceutical product, if the fact of a prescription is considered to be the unit of output (independent from the rest of the health care pathway). If the unit of output is considered to be the health care pathway, then the exact type of chemical is not needed: all that would be used is the cost of the prescription in order to give the health care pathway the correct (total) costs weight.
  • Price, adjustment amount, reimbursement cost, national adjustment, patient contribution, dispensing fee value, and subsidy value These variables could be used as part of a set of information used in constructing the relative weight. Information on the patient co-payment (and other contributions to total cost) would also be needed to ensure that the relative weight is correctly specified.
  • Quantity, quantity units, prescribed quantity, quantity dispensed, quantity prescribed, frequency, dispensing supplied, dose, days supply, daily dose, and weight These variables identify the quantity of formulation numbers (these are standardised according to chemical entity) and formulation strength, and could be used to ensure that the correct price / volume breakdown is made.
  • Repeat sequence number This variable identifies the number of repeat prescriptions (number 1 is a prescription given as part of contact with a doctor, 2+ are the repeats). This variable could be used to identify which prescriptions can be aggregated if the unit of output is defined to be the health care pathway. For chronic conditions, where the unit of output may be along the lines of ‘a patient with chronic condition X managed over time period Y’, this variable would allow aggregation of prescription costs as part of estimating total costs for that particular unit of output.

Level of disaggregation

Pharms includes a number of variables which capture information on the characteristics which might be considered to be important to the consumer. These include: age at dispensing, formulation ID, quantity, quantity units, prescribed quantity, quantity dispensed, quantity prescribed, frequency, dispensing supplied, dose, days supply, daily dose, and weight.

Overlaps / duplication with other sources

None within the National Collections.

Access to dataset

The MoH has access to the dataset for analytical purposes.

Corresponding weights

Pharms includes complete information for constructing relative weights.

top

8.1.9 Other sources

Other information is collected outside of the framework of the National Collections. This sub-section describes the most important of these, in terms of comprehensiveness of health system activity.

Maternity services

A bespoke collection for monitoring funding of maternity events, administered by the Sector Accountability and Funding Directorate within the Ministry of Health.

Disability support services

There are two sources of information on funding of disability support services: the Contract Management System (CMS) and the Client Claim Processing System (CCPS), which together form the basis for managing the flow of finances from purchasers (either the Ministry or DHBs) to providers (care homes and so on). Both sources hold information on activity (although at different levels of aggregation, and on financing (prices)).

CMS is complete for the type of services for which it is the system for ensuring flow of finance. It holds information on bulk purchases, and is therefore at an aggregate level (in many cases, for example, the unit of measurement is the ‘contract’).

CCPS, by contrast, is thought to be incomplete. It holds information at the event level.

Many disability support services are purchased on the basis of inputs, for example bed days, so care needs to be taken when configuring the unit of output in a measure of productivity.

Care needs to be taken with Disability support services, as these are typically a joined-up mix of different services to parts of the population that need health as well as social care. The health and social care boundary is fraught with definitional difficulties, with much effort made internationally by the OECD and others to try to tighten these definitions. Currently, there exists no agreed consensus on where the exact borderline lies, but there is guidance to help, in the OECD’s A system of health accounts (OECD 2001b), and in the UK and Eurostat’s SHA guidance (ONS 2005).

Mental health services

DHBs provide the Ministry with information on mental health services that is distinct and independent from the information collected for NMDS and NNPAC.

Many mental health services are purchased on the basis of inputs, for example bed days, so care needs to be taken when configuring the unit of output in a measure of productivity.

Public health services purchased from hospitals

The Ministry purchases some public health services directly from hospitals. Activity is not particularly well captured. For example, there is no data warehouse designed for this purpose.

Price Volume Schedules

Price Volume Schedules (PVS) are part of the way in which District Health Boards inform the Ministry of Health about their expenditure and activity plans to meet local demand for services and government priorities for the provision of health care services. The schedules contain information at an aggregated level on the volume of services to be purchased and/or provided by the DHB, and the price of these services. The level of aggregation is fixed, such that all DHBs report at the same level.

The DHBs tend to publish the information in their PVS as part of their Annual Report. See, for example (WCDHB 2009) and (MCDHB 2009).

Care needs to be taken in using the information in the PVS for the purpose of measuring productivity change, as much of the information they contain relates only to planned or forecast activity, rather than actual outturn (comparison against outturn is carried out at a much more aggregated level). Also, the unit of measurement is casemix-adjusted. There is no reporting of raw number of procedures, so it is not possible to undo the case-mix adjustment without access to raw data.

top

8.2 Health care: output quality

The development of system-wide level measures of change in the quality of health care provided in New Zealand at a system-wide level is in its infancy, as is the case for most other countries. That is not to say that there is a dearth of information on the quality of individual services: it is at the level of the entire health system that information is lacking. This can be put down to two main reasons. Firstly, no overarching model exists for aggregating the various indicators of quality. Secondly, there is yet to be discussion and agreement as to what indicators of quality should be taken into account for a system-wide level measure of health care quality.

Recommendation H10

Given the development infancy of system-level measures of change in the quality of health care provided in New Zealand, and that until there is broad discussion and agreement on how to construct such measures and combine these with the existing quantity measures, care should be taken in presenting such information.

This sub-section describes the main indicators which are already considered useful for understanding how the system-wide level of health care quality may be changing over time in the UK. Guidance on what might be considered useful has been taken from the Atkinson review, which has met with considerable support from other countries and from international institutions. However, this report recommends that New Zealand considers for itself how to construct system-wide measures of health care quality change. 

 

Recommendation H10

Given the development infancy of system-level measures of change in the quality of health care provided in New Zealand, and that until there is broad discussion and agreement on how to construct such measures and combine these with the existing quantity measures, care should be taken in presenting such information.

This sub-section describes the main indicators which are already considered useful for understanding how the system-wide level of health care quality may be changing over time in the UK. Guidance on what might be considered useful has been taken from the Atkinson review, which has met with considerable support from other countries and from international institutions. However, this report recommends that New Zealand considers for itself how to construct system-wide measures of health care quality change. 

top

8.2.1 Health care quality measures in the Atkinson Review

The Atkinson review describes two dimensions of quality which are important: health care effectiveness and patient experience. As with characteristics for any set of goods and services, the relative importance of these dimensions, and of individual indicators within these dimensions, depends on the individual good or service. For example, the ratio of importance of health care effectiveness (eg survival) to patient experience (eg politeness of health care staff) is probably of the order of 1:0 in the case of patients with acute myocardial infarction (AMI or heart attack). This will not be the case for patients presenting to their local GP with routine influenza symptoms (although clearly in some cases there is a risk of mortality, in which case the ratio should not be 0:1, but may not be too far off).

The model used in the UK combines measures of change in quantity and the various aspects of quality in a fairly simple way: the model is multiplicative and assumes that the relative importance of a 1 percentage point change in all of the indicators are of equal importance. Therefore, a 1 percent increase in the quantity of a particular operation, and a 1 percent increase in the survival rate for that operation, and a 1 percent increase in the effectiveness of that operation, means that total measured output change will be 3.0301 percent (1.01 * 1.01 * 1.01).

An improved model would be very data intensive, requiring information on the relative importance of each of the components of total output change, as well as empirical studies about the way in which the components interact with each other. While some work to produce better models has been carried out for individual diseases, there is as yet very little work available at the level of the aggregate health care system.

top

Health care effectiveness: survival, health care effectiveness, and waiting times

In the UK, a model for inpatient and day care in hospitals has been adopted, which combines the effects of three indicators of quality change: survival, health care effectiveness (for those that survive), and the health care effects stemming from changes in waiting times (changes in waiting times are considered to have both health care effects as well as patient experience effects: examples of each include the importance of waiting time for life-threatening conditions and the convenience of patients having outpatient appointments at times that suit them rather than the outpatient clinic). The specification of the model is complex and is not reproduced here. The details can be found in York’s paper Developing new approaches to measuring NHS outputs and activity (York 2005).

The basic specification of the model is that it is multiplicative: relative changes in each of the components are multiplied together on the assumption that each component is equally important. The model is applied at the level of individual DRGs, so in theory it takes into account differential changes in quantity and the different dimensions of quality for different classes of disease (although, in practice, little empirical evidence is available).

The model also takes into account the fact that, for some types of disease, there may be a very low or very high chance of death. In such cases, the model makes no quality adjustment.

The survival indicator takes the form of survival within 30 days of discharge, which requires linkage between heath and population, and vital events databases, which are held by different organisations (in both the UK and New Zealand, by the health and statistical authorities respectively).

The waiting times indicator actually measures change in waiting times at the 80th percentile, in order to strip out the effects of erroneous reporting of extremely high waiting times.

There has been discussion in the UK about refining this particular quality measure to cover only those diseases for which hospital admission can make a difference to the health status of the patient; that is, to confine the measure to avoidable or amenable mortality, but this has not yet been enacted, mainly due to difficulties in mapping the DRG and avoidable or amenable mortality classifications.

Information on survival after hospital admission is available in New Zealand: the Ministry regularly updates a linkage between hospital activity data (held in the NMDS) and population and vital events data (held by Statistics NZ). This linkage can be carried out at any level of aggregation, given the detailed information that exists in NMDS and the appearance of the NHI on both the NMDS and in the vital events register.

Avoidable and amenable mortality information is available in New Zealand, through the NZHIS.

Information on the health effects of hospital treatment is not routinely collected in New Zealand, as is the case in the UK and many other countries. There is growing interest in the UK in collecting information on this, not least for understanding health care performance at the system-wide level, but mainly because of the need to understand the efficiency and effectiveness of what hospitals are doing. There is, therefore, similar interest growing in New Zealand. It is worthwhile ensuring that in discussions of data uses, that Statistics NZ formally lodges its interest in such information.

Recommendation H12

Statistics NZ should formally register its interest in information on the effectiveness of hospital treatment, as part of an information suite that could be used in measuring health care output at the national level.

Information similar to waiting times is available, although there are issues with interpretation. The source is the national patient booking system. The information available is not on waiting per se, but on booked appointments. Further work will be needed to understand the nature of information from this system.

top

Management of chronic conditions in primary care

In the UK, a payment and reward system for General Practitioners exists, entitled the Quality and Outcomes Framework (QoF). A broad range of extra payments is available to incentivise good clinical practice, and in order to ensure appropriate payments are made, a great deal of information is collected. Some of the indicators used in QoF are output-related, such as the extent to which some chronic conditions are effectively managed. Increases in the proportion of patients whose conditions are effectively managed are used as quality adjustments.

Patient satisfaction

In the UK, there are a number of surveys of patients’ experience, each of which is specific to the type of health care received; including, for example, hospital inpatient acute care, mental health services, and so on. Not every survey is carried out for every year (indeed some of the surveys are very occasional), and there are changes in the nature of the questions asked between similar surveys over time. Nevertheless, there is a subset of information which can be used to track changes over time in patient experience. Where this is possible, indicators are constructed that identify the percentage increase in satisfaction over time, and are used in the UK health care output model.

The Ministry collects some information on patient experience through patient surveys after receiving health care. These are described in, for example, the New Zealand Medical journal. See for example http://www.nzma.org.nz/journal/122-1300/3738/.

top

8.3 Health care: output quantity weights

8.3.1 Weighted Inlier Equivalent Separations (WIES)

The Ministry of Health operates a casemix funding system, at least in part, for funding its hospitals. In simple terms, casemix funding systems require historical information on activity carried out in hospitals and corresponding historical costs, in order to construct reimbursement ‘prices’ for current (and future) hospital activity.

Although there is much information available on activities carried out in New Zealand hospitals, until recently there has been very little or no actual New Zealand hospital costs collected: 2008/09 will be the first year for which solely New Zealand cost information will be used in the New Zealand casemix funding system. For previous years, the reimbursement prices for hospital activity in New Zealand have been based on Victorian (Australia) hospital costs.

As the Ministry of Health has adopted the overarching structure of the Victorian casemix funding system, it is important to understand the system in Victoria in order to understand the system in New Zealand.

The calculation of WIES was originally, and continues to be, a significant part of the casemix funding system in Victoria, Australia. This casemix funding system was introduced initially by the state of Victoria on 1 July 1993, as part of a fundamental change in the way in which hospital care was funded: prior to this date, hospital treatment was funded on a historical basis and subject to detailed input controls; from this date, the majority of hospital treatment was funded on a casemix basis. The Victorian casemix funding system involves the following steps (source: State Government of Victoria, Australia):

  1. diagnoses for each patient are recorded and coded to a DRG
  2. each DRG has a particular ‘weighting’ set around a value of 1. The weighting is derived through annual costing studies that compare the relative resource consumption of each DRG against all others
  3. the aggregate number of DRGs in any time period, multiplied by the weighting of each, results in a number called a weighted separation (a separation is a discharged patient event)
  4. the system recognises outliers when the length of stay is abnormally long, or abnormally short – according to agreed statistical parameters. Short stay outliers receive a reduced payment and long stay outliers an increased payment. These payments can be converted into the equivalent of DRG weights. This conversion collapses all DRG payments into a single number – the Weighted Inlier Equivalent Separation, or WIES.
  5. WIES are then multiplied by the price (set annually for each grouping of similar hospitals) per unit of WIES (the price paid for a notional DRG with a weighting of 1) to determine the funding available within any time period.

The major modification that the Ministry of Health has made to the Victorian system is that where New Zealand actual data are available, these replace the Victorian data (for example, New Zealand actual costs have been collected for use in the 2008/09 year, thus replacing the Victorian weightings calculated in steps 1, and New Zealand specific thresholds have been introduced for identifying outliers in step 4).

The WIES from the New Zealand casemix funding system is calculated as follows:

(v)

where j denotes an activity within a DRG, and the asterisk (*) denotes Victorian data.

top

Currently, the MoH methodology for calculating volume change over time involves comparing the sum of the WIES for each year, as follows:


(vi)

Note: The extra term has been introduced to account for the differing number of activities between DRGs. Rearranging equation (vi) gives:

top

equation for 8.3
(vii)

The first term is the change in the raw number of Victorian discharges.

The second term is the inverse of change in total Victorian expenditure.

The third term is change in total Victorian expenditure, the numerator and denominator of which have been modified with the ratio of New Zealand to Victorian number of raw discharges at the level of DRGs.

If the ratio of raw discharges within DRGs remains constant, the third term is simply change in Victorian expenditure, and thus this and the second term cancel, leaving only the first term.

Where only New Zealand data are used in calculation, the second and third term cancel, again leaving only the first term.

This is a simplified model. There are several complications, but these do not perturb the basic finding that the comparison of the sum of the WIES for a pair of years leads to a comparison of the number of raw discharges:

  • Costs are also calculated on the basis of outlier / inlier status and LOS. The model would cope with this extra complexity by changing the definition of j, which would denote the combination of DRG, inlier / outlier status, and LOS, thus leading to the calculation of a much larger number of WIES, one for each combination.
  • The NZ DG classification is modified from the Victorian one. This would complicate the mathematical presentation immensely, and introduce an extra (set of) term(s) to deal with the differences.
  • The coverage of hospital activity in Victoria for calculating the WIES is different from the coverage of hospital activity in NZ to which the WIES are applied. Again, this would complicate the mathematical presentation immensely, and introduce an extra (set of) term(s) to deal with the difference in coverage.

top

When multiplied by the price in NZ$ of the average treatment in New Zealand, these weights could be used as the basis for relative weights for the growth rates in hospital inpatient and day care activity. However, there are some issues that need to be explored.

  • The weights are designed for the purpose of reimbursing hospitals for Ministry funded activity. The relative costs / prices for activity funded by other sources, including from the Accident Compensation Corporation (ACC) and private sector payments, may not be the same as those implied by the WIES weights, and thus the WIES weights may be a biased measure of the relative importance of the different types of activity. Further study of the relationship between the WIES weights and total hospital costs, including all sources of funding, may confirm or reject the existence of any bias.
  • As the historical weights are based on Victorian activity and costs, when applied to New Zealand activity, they do not average to 1. Simple pro-rating will resolve this issue.

Recommendation H13

Statistics NZ and the Ministry of Health should study the relationship between the WIES weights and total hospital costs, including all sources of funding, to confirm whether use of WIES weights, as the measure of relative importance of different types of hospital inpatient and day care activities, introduces any bias. 

top

8.4 Health care: inputs

8.4.1 Household Labour Force Survey

Overview

The Household Labour Force Survey (HLFS) is a major quarterly survey run by Statistics NZ. It collects information on the labour market participation of those living in private households in New Zealand.

Coverage

The sampling frame for the HLFS is the civilian, non-institutionalised, usually-resident New Zealand population aged 15 years and over.

Timing (frequency, time reference, length of time series)

The survey is collected quarterly.

Use in a measure of productivity

The survey collects most of the information that would be needed to construct a measure of labour inputs to production for the health sector, including number of personnel, hours worked, salaries, as well as information that would help distinguish between the different types of labour input; for example, occupation.

The results of the survey may also be useful when combined with information from other sources. For example, if other sources can only furnish information on contracted hours, then the HLFS could provide information on the relationship between contracted and actual hours worked.

Known issues

Sampling error would be a potential issue, especially if a fine level of detail is required; for example, in differentiating between different types of labour. One way around this may be to combine estimates over a number of periods. However, as well as decreasing the confidence intervals for the estimates of interest, this approach would smooth any changes over time, and would therefore be less sensitive to real change.

The information on income is limited to wages and salaries, and self-employment income, and does not collect information on other parts of employment-related compensation (which would primarily be known by the employer rather than employee). Wages and salaries, though, may be a reasonable proxy for total compensation.

The HLFS does not collect any explicit information that would allow a distinction to be made between those working in the public and private sectors.

Key variablestop

  • Number of those in employment, number of those in self employment, industry, occupation, actual hours worked, education, participation in formal study.

Level of disaggregation

The source includes a number of variables which capture information on those characteristics of the medical workforce that might be used to differentiate between different kinds of labour input.

Overlaps / duplication with other sources

Much of the information collected in the HLFS on health care labour is available elsewhere.

Access to dataset

Statistics NZ collects and processes the raw data. Subject to confidentiality assurance, data are available.

Corresponding weights

The New Zealand Income Survey collects information on wages and salaries, and self employment income, as a supplement to the HLFS in the June quarter.

top

8.4.2 Medical Council’s Workforce statistics

Overview

The Medical Council of New Zealand collects and publishes information in the New Zealand Medical Workforce (Med 2008). The information is collected as part of the annual renewal practicing certificates under the Health Practitioners Competence Assurance Act 2003. Early renewals (and information collections) were carried out under the Medical Practitioners Act 1995.

Coverage

Renewals are annual, and are staggered over the year. Depending on the birth date of the doctor, the renewal dates are November, February, May, or August.

In 2008, the response rate was 87 percent.

Changes in the Council’s registration policies has meant that the sampling frame now includes some doctors who previously held temporary registration and would have been excluded. The sampling frame does not include doctors registered for specific short-term purposes.

It is thought that the inclusion of these doctors who previously held temporary registration in the sampling frame has lead to a drop in the response rate, due, for example, to the temporarily registered doctors having left New Zealand by the time the survey is posted.

The results of the survey as published present information on active doctors: the definition of ‘active’ is that the doctor must have worked for four or more hours a week. The definition of Full time equivalent (FTE) is 40 hours per week.

Timing (frequency, time reference, length of time series)

The information is collected annually.

Two major changes have been introduced since 1992:

  • Repeat prescriptions were introduced in 1996; and
  • The major drug key changed from Medicode to Pharmacode in 1998.

Use in a measure of productivity

This source of information on doctors in New Zealand would potentially be a very powerful booster alongside more general surveys of the workforce, given that the sampling frame covers all doctors in New Zealand, the survey is sent to all doctors, and that the response rate is 87 percent.

Known issues

The changeover in law to the Health Practitioners Competence Assurance Act 2003 from the Medical Practitioners Act 1995 has meant that there are changes to some of the survey elements , including the terminology used.

Key variablestop

  • Information collected in the survey is combined with information already held on the Council’s databases, to avoid repeatedly asking for the same information; for example, sex, age, registration date, and country of graduation.
  • Information available from the survey (and associated databases) includes: region; length of service, country of registration; age; sex; work type (eg primary care or house officer); vocational scope (eg anaesthesia, emergency, or ophthalmology); hours worked; and type of employer.

Level of disaggregation

The source includes a number of variables which capture information on those characteristics of the medical workforce that might be used to differentiate between different kinds of labour input. These include: length of service, work type, vocational scope, and type of employer.

Overlaps / duplication with other sources

Other surveys include information about doctors employed in the health care industry.

Access to dataset

The MoH has access to the dataset for analytical purposes.

Corresponding weights

Other sources include information on compensation of employment; however, they may not match the level of disaggregation offered in this source.

top

8.4.3 Ministry of Health’s National Asset Management Plan

Overview

The Ministry, together with the health care sector, has produced two National Asset Management Plans, the first in 2006 and the latest in 2009 (the latter is still in draft form). The Plan has been designed to help the sector understand what assets it has at its disposable, as well as to help it prioritise asset planning in the future.

Coverage

Information is collected from all DHBs in New Zealand.

Timing (frequency, time reference, length of time series)

There have been two Plans produced, the first in 2006 and the latest in 2006 (still in draft at the time of writing this report).

Use in a measure of productivity

The Plan is based on a great deal of information collected from DHBs on the current status of their asset base, which could be used in conjunction with other available information as the raw data for a health care industry-focused PIM.

Known issues

The information collected relates to assets held by the DHBs, and does not cover assets held by private sector actors. It therefore covers the majority of secondary and tertiary care, but a much smaller proportion of the primary care sector.

Key variablestop

  • Type of asset, replacement cost, functionality, condition.

Level of disaggregation

The source includes variables such as type of asset, which would be the main way of disaggregating the information on assets for the PIM.

Overlaps / duplication with other sources

The information collected for the Plan may be from the same original sources as information collected by Statistics NZ. This would need to be looked into.

Access to dataset

The Ministry holds this information. Given its commitment to providing information on capital assets to Statistics NZ for the PIM, there should be no problem gaining access to other such datasets.

Corresponding weights

See above.

top

8.5 Health care: complementary indicators

This sub-section sets out a number of complimentary indicators that might be useful in helping to interpret health care output, inputs, and productivity indicators, as discussed in section 5.4.3. The indicators presented are those presented in the UK's Public service productivity: health care articles (ONS 2008), and are: average length of stay in hospital; elective day case rate; emergency readmission rates; and number of operations cancelled.

top

8.5.1 Average length of stay in hospital

Reducing the length of stay in hospital has been a major driver of reduced costs and increased productivity in the English NHS: more patients can be treated with the same number of hospital beds and other fixed resources. The NHS Institute for Innovation and Improvement has estimated that £975m ($2.2bn) could be saved each year by reducing the average length of stay in hospitals to the level of the top 25 percent of hospitals (if all else remains unchanged)

Care needs to be taken in interpreting movements in the average length of stay in hospital alongside productivity performance, as there are other factors beyond productivity change that could explain changes in the average length of stay in hospital, such as a change in casemix. It is possible that additional resources are being used to reduce length of stay – for example, more clinical staff on duty at weekends – outweighing the savings being made.

top

8.5.2 Elective day case rate

As well as reducing the costs to the health service, increasing the day case rate plays a part in providing timely treatment, in reducing the risk of cross infection, and in reducing the number of procedures cancelled (BMJ 2005). Treatment by day case surgery is also seen to have a positive quality of life effect for the patient. This is because the procedure is likely to have a shorter waiting time; patients can return home the same day, which means an earlier return to normal activities; and patients can potentially receive care that is better suited to their needs. Treating patients as day cases instead of in-patients would be expected to reduce required inputs

top

8.5.3 Emergency readmission rates

Emergency readmissions are generally unlikely to be part of the patient’s originally planned treatment, and some may be avoidable (NCHOD 2005). Readmission rates are often used as a measure of the quality of care received by patients in health care systems (HSJ 2004).

There are a number of factors that could explain change in the number of emergency readmissions. For example, hospitals could be dealing with more complex cases, some patients may have more severe symptoms, or hospitals could be discharging some patients too quickly after treatment. But in some cases, readmission may be part of a deliberate plan, agreed between clinicians and patients, to allow patients to return home earlier on the understanding that they will be readmitted immediately if needed.

Care therefore needs to be taken in interpreting changes in emergency admission as a complimentary indicator alongside a measure of productivity change.

top

8.5.4 Number of operations cancelled

The number of operations cancelled at the last minute for non-clinical reasons can show how efficiently a health system uses resources. From the perspective of a patient, having an operation cancelled at the last minute is far from desirable.

Again, care should be taken when interpreting changes in number of operations cancelled as a complimentary indicator alongside productivity change. Many factors could influence changes in the number of operations cancelled at the last minute, some of which are outside the control of a health system/provider. These include:

  • how effectively hospitals manage their resources and appointments
  • the commitment of patients in keeping to their appointments
  • the need to divert resources to deal with unexpected pressures from emergency admissions.

top

8.5.5 Amenable mortality

Three types of mortality can be described:

  • Amenable mortality – deaths occurring before age 75 from causes that are considered amenable to medical intervention.
  • Preventable mortality – deaths occurring before age 75 from causes that are considered to be preventable through a) individual behaviour, and/or b) public health measures limiting individual exposure to harmful substances/conditions.
  • Unavoidable mortality – deaths occurring before age 75 from causes that are considered a) not amenable to medical intervention and b) not preventable through changes in individual behaviour/public health measures.

A comparison of the trend in mortality from these three types of cause (repeated in ONS 2008) has shown a considerable decrease in mortality from causes amenable to medical intervention, whereas mortality from causes considered to be unavoidable decreased only modestly. One interpretation of this is that medical interventions have contributed positively to the reduction in avoidable mortality.

However, there are uncertainties about the attribution of the role played by the health service in reducing amenable mortality, with further study of the attribution needed.

  • Share this page to Facebook
  • Share this page to Twitter
  • Share this page to Google+
Top
  • Share this page to Facebook
  • Share this page to Twitter
  • Share this page to Google+