  |
|
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| Journal of the New Zealand Medical Association,
21-July-2006, Vol 119 No 1238
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Will
a web-based cardiovascular disease (CVD) risk assessment programme increase the
assessment of CVD risk factors for Māori?
Robyn Whittaker, Dale Bramley, Sue Wells, Alistair Stewart,
Vanessa Selak, Sue Furness, Natasha Rafter, Paul Roseman, Rod Jackson
Cardiovascular
disease (CVD) is the leading cause of death for
Māori.1
Despite priority being placed on CVD by the health system in New Zealand (NZ),
inequalities in cardiovascular health outcomes have a negative impact on
Māori.2
Māori also have high prevalence of many cardiovascular risk factors, such
as smoking,3 diabetes, high cholesterol, and elevated blood pressure.4
Waitemata District Health Board (WDHB), in concordance with
the Ministry of Health’s national priority areas,5 has a focus on CVD
prevention and on reducing
inequalities.6 Current
New Zealand guidelines recommend cardiovascular risk assessment and management
for all Māori men over 35 years of age and Māori women over 45 years
of age—and 10 years later for people of non-Māori, non-Pacific,
non-Indian subcontinent ethnicity.7
A web-based tool to facilitate risk assessment and
management in primary care (PREDICT-CVD) has been developed in New Zealand and
implemented using the brand name “Prompt” in ProCare Health Limited
(a group of three Primary Health Organisations (PHOs) under one Management
Services Organisation). This PREDICT–CVD programme is integrated into the
patient management system to allow systematic cardiovascular risk assessment and
provide within-seconds evidence-based patient-tailored decision support
according to the national guidelines for the management of cardiovascular risk.
Prior to supporting the implementation of such a tool,
Waitemata DHB sought evidence that the use of PREDICT-CVD would not
adversely impact
inequalities. Therefore a retrospective before-after study was designed to
investigate the effect of PREDICT-CVD on documentation of CVD risk and risk
factors for Māori and non-Māori.
MethodsEligible general practitioners (GPs) were those who had
used MedTech32 electronic medical records for at least 1 year, had installed
PREDICT-CVD in their practice prior to May 2004, and used systems categorising
patients according to their registered GP.
They were invited by ProCare Health Ltd to participate
in the study. Researchers conducted electronic queries to create lists of
patients meeting the NZ guideline criteria for cardiovascular risk assessment
(Māori/Pacific/Indian
subcontinent men aged 35 years and over; Māori/Pacific/Indian subcontinent
women aged 45 years and over—and 10 years later respectively for others),
who had visited an eligible GP within a 4-week period that was either 1 month
after the GP first used PREDICT or the same 4-week period 1 year earlier
(pre-PREDICT period).
These time periods ranged from August 2001 to June
2004, with considerable overlap between pre-PREDICT and post-PREDICT
installation time periods. Both time periods were audited retrospectively, and
GPs did not know in either period that they would be audited.
The study was designed to provide adequate explanatory
power for Māori.
Therefore all Māori patients identified, and a randomly selected 15% sample
of non-Māori patients identified, were included in the audit of electronic
medical records (EMRs).
Patients
whose ethnicity was documented as NZ Māori or equivalent (Māori, M, or
New Zealand Health Information Service [NZHIS] code 21) in the EMR were included
as Māori. Where no ethnicity was recorded, patients were assumed to be
non-Māori.
Trained and experienced audit nurses conducted the EMR
audits in the general practices, recording any cardiovascular risk notation
(range, %, text description), and any documentation of smoking status (smoker,
non-smoker, past smoker), cholesterol level (total cholesterol:HDL ratio or
total cholesterol), blood pressure measurements, and diabetes status (type 1 or
type 2 diabetes stated, diabetes documented as absent; or no documented
diagnosis but evidence of impaired glucose tolerance, raised HbA1c, or
prescriptions for insulin/test strips/oral hypoglycaemics).
All analyses were conducted
using SAS statistical
software Version 9.1. For the outcomes of interest, the proportions for the
total population were derived from Māori and non-Māori sampling
populations. To calculate odds ratios together with 95% confidence intervals, a
multivariate analysis was conducted using a mixed logistic regression
model in which GPs were regarded as random effects and all other variables
regarded as fixed effects.
The model included the practice that the GP worked in
and patient characteristics that may influence GP risk-assessment behaviour:
age, gender, ethnicity, presence of existing cardiovascular disease, diabetes,
High Use Health Card (HUHC – for those with medical conditions requiring
frequent GP visits), and Community Services Card (CSC – for low income
families). To assess if documentation of risk or risk factors differed by
ethnicity after the implementation of PREDICT, an interaction term was used.
The PREDICT-CVD Evaluation Study gained ethics approval
from the Auckland Regional Ethics Committee (AKY/04/07/185).
ResultsOf the 107 eligible GPs, 84 (78.5%) consented to participate
(4 were absent from their practices, 1 could not be contacted, and 18 declined).
Four practices were unable to supply data, thus leaving 80 contributing GPs.8
A total of
3564 audits were
conducted (1680 in the pre-PREDICT period and 1884 in the post-installation
period). Māori participants made up 28.2% (n=474) of audited EMRs in the
pre-PREDICT period, and 25.7% (n=484) in the post-PREDICT period.
The
audited Māori participants were significantly different from the
non-Māori participants in all parameters evaluated. Māori participants
were younger (due to the different sampling criteria from the screening
guidelines) with very few aged over 74 years; were more likely to be
male, have diabetes, and a CSC; and were less likely to have a HUHC and
previous CVD, than
non-Māori
participants.
The
characteristics of Māori participants did not differ greatly between the
two time periods. The only significant difference among non-Māori
participants was a higher level of previous CVD in the post-PREDICT period (25%
of participants compared to 21% in the pre-PREDICT period, Chi-squared=5.65,
p=0.0175).
Table
1. Characteristics of Māori and non-Māori audited populations
Prior
to PREDICT installation, Māori had slightly higher recording of
cardiovascular risk than non-Māori (in 3.2% vs 2.8% of EMRs). Where the EMR
contained information on all the risk factors necessary to conduct a risk
assessment but overall risk was not documented, the GP may have used a
paper-based risk assessment chart to calculate CVD risk but not documented it in
the EMR. The baseline proportion of EMRs with
either risk documented
or all
necessary risk factors
documented was 12.4% in Māori and 9% in non-Māori. (Figure
1)
There
was an increase in the recording of risk after the installation of PREDICT for
both Māori and non-Māori. This increase appears greater in Māori
participants (from 3.2% of EMRs to 14.7% of EMRs) than in non-Māori (from
2.8% to 10.5% of EMRs), although it was not statistically significant.
When the recording of CVD risk factors was included,
documentation of risk or risk factors increased from 12.4% to 24.0% of EMRs for
Māori and 9.0% to
17.6% for non-Māori, but again this increase was not statistically
different between Māori and non-Māori.
Figure
1. Proportion of EMRs with CVD risk, and risk factors, documented pre- and
post-PREDICT by Māori and non-Māori
 The
recording of both smoking and diabetes status was higher for Māori than
non-Māori in both periods. The recording of smoking status for Māori
increased from 49.6% of EMRs prior to PREDICT to 59.3% in the post PREDICT
period. Recording of diabetes status also increased from 21.5% to 23.3% (Table
2). The increase in the recording of all risk factors from pre- to
post-installation was not statistically different for Māori compared to
non-Māori.
Table 2. Proportion of EMRs with risk factors
documented
*Model includes: age, gender,
presence of CVD, diabetes, HUHC, CSC.
DiscussionAs District Health Boards have a legislative mandate to
reduce inequalities
and improve Māori health, it was considered vital for Waitemata DHB to
assess the potential effect of a comprehensive cardiovascular risk assessment
and management programme on inequalities prior to the introduction of such a
programme.
Therefore an
evaluation study was designed specifically to have adequate explanatory power
for Māori.
The use of this methodology could be considered an
innovation by the DHB as it attempts to find ways to increase health service
responsiveness to the need to reduce health inequalities. This type of study
methodology could also be used in the future by other DHBs in order to evaluate
new or existing services and their impact on inequalities.
The results of this study are encouraging in that risk
factor documentation
and risk assessment
increased similarly for both Māori and non-Māori after the
installation of the PREDICT-CVD risk assessment tool. It appears that this type
of tool will not increase inequalities and, if applied in an appropriate manner,
could potentially lead to a reduction in inequalities. However,
introduction of the risk assessment tool alone is not enough as evidenced by the
low overall level of documented risk assessment.
In future we recommend introduction be accompanied by a
comprehensive implementation programme to ensure that the entire target
population is assessed. Also, the ensuing management of risk still needs to be
followed up to ensure that risk assessment does lead to better treatment and
health outcomes, and to avoid differences in treatment by ethnicity.
It
is of interest that the recording of cardiovascular risk and risk factors for
Māori was so low in this population given the demonstrated high prevalence
of cardiovascular risk factors in Māori in other studies. For example, the
New Zealand Health Survey (2002/03) found that 23.7% of Māori adults had
been told they had high blood pressure (vs 17.6% European/Other) and 15.9% high
cholesterol (vs 14.6%
European/Other).9
Also CVD is known to occur at an earlier age in Māori, as reflected in the
cardiovascular risk guidelines recommendation to screen this group 10 years
earlier. Should Māori patients be prioritised for risk assessment and risk
factor documentation, the health gains are likely to be large.
The documentation of smoking and
diabetes status were
higher in Māori than non-Māori. Just under one in every two Māori
are smokers compared to one in every five non-Māori,
3 and the prevalence
of diabetes is 6.7% in Māori compared to 2.4% in European/Others. 9
Therefore documentation is likely to be higher
among
Māori if GPs are more likely to record a positive finding than a
negative one.
The
audited sample contained very few older Māori. This is likely to be
influenced by the fact that Māori live (on average) 8–10 years less
than
non-Māori.10
There were also low numbers of Māori participants with a HUHC, which is
surprising given the high prevalence of diabetes and other chronic conditions in
Māori. This could reflect inadequate targeting of this tool.
There was a lower rate of
documented previous
CVD in Māori than non-Māori in this group. The reasons for this are
unknown but should be explored further.
Limitations of the
study—Māori
are known to be proportionately under-enrolled in PHOs in some areas 11
therefore it is possible that
the ProCare Health Ltd
enrolled population may not be fully representative of the Māori population
of the Auckland region. Differences in access to primary care for Māori,
even when enrolled, may also have affected selection of participants.
Also, the definition of ethnicity may lead to classification
error, as there was considerable variability across practices in the recording
of ethnicity (NZHIS codes, other codes, free text). In the ProCare Health
Ltd-enrolled population at the time of this study, 12.3% of patients had no
ethnicity recorded [personal communication Ken Leech, ProCare Health Ltd, 2005],
and an ensuing audit of ethnicity coding found that less than 2% of audited
patients had more than one ethnicity recorded.12
In this study 7.2% of audited
EMRs had no ethnicity
stated and these were coded as non-Māori, which may have resulted in
undercounting of Māori participants. It is not known whether the documented
ethnicity was self-identified by the patient or assigned by the practice.
Suggestions for the
future—The implementation of the PREDICT-CVD tool should not occur
in isolation without a comprehensive quality-driven programme. Aspects of such a
programme could include: education for primary care staff concerning the
importance of documenting disease risk factors, even if these are found to be
absent/negative; active recalling of patients meeting the NZ guideline criteria
from PHO-registered population lists; a standard approach to the documentation
of ethnicity in primary care; automatic prompts
for EMRs of patients
who meet the criteria for risk assessment according to the New Zealand
guidelines; and targeting risk assessment for those most in need (in particular
Māori and Pacific). It should also include taking risk assessment into
community settings to provide access for those not currently engaged with
primary care.
An ongoing challenge for DHBs is the requirement to explore
and implement healthcare practices that will lead to a reduction in health
inequalities.
Author information:
Robyn Whittaker, Public Health Physician, Health Gain Team, Waitemata District
Health Board; Dale Bramley, Manager of Health Gain, Waitemata District Health
Board; Sue Wells, Senior Lecturer – Clinical Epidemiology, Epidemiology
and Biostatistics, University of Auckland; Alistair Stewart, Biostatistician
Epidemiology and Biostatistics, University of Auckland; Vanessa Selak, Public
Health Medicine Registrar, Epidemiology and Biostatistics, University of
Auckland; Sue Furness, Project Manager, Epidemiology and Biostatistics,
University of Auckland; Natasha Rafter, Public Health Medicine Registrar, Health
Gain Team, Waitemata District Health Board; Paul Roseman, Manager Design &
Development, ProCare Health Ltd; Rod Jackson, Professor of Epidemiology,
Epidemiology and Biostatistics, University of Auckland; Auckland
Acknowledgements:
This study was supported by funds from Waitemata District Health Board
and The Future Forum while Sue Wells is the recipient of a National Heart
Foundation Research Fellowship.
We also thank Elaine Horn, Kate Moodabe, and all the GPs
from ProCare Health Ltd for participating in the study; their practice teams for
making us feel welcome; the audit nurses and staff from the Diabetes Project
Trust; and Waitemata DHB Cardiovascular Technical Advisory Group for
recommending we undertake the study.
PREDICT-CVD was developed by the University of Auckland and
Enigma Publishing Ltd in collaboration with ProCare Health Ltd, Counties Manukau
District Health Board, Ministry of Health, National Heart Foundation, New
Zealand Guidelines Group, and MedTech Global Ltd.
Correspondence: Dr
Dale Bramley, Manager of Health Gain, Waitemata District Health Board, Private
Bag 93-503, Takapuna, Auckland. Fax: (09) 441 8957; email: dale.bramley@waitematadhb.govt.nz
References:
|
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| | Current
issue | Search journal |
Archived issues | Classifieds
| Hotline (free ads) Subscribe | Contribute | Advertise | Contact Us | Copyright | Other Journals |
|