Health Provider Use of PREDICT Electronic Decision Support Within Primary Care

1
Health Provider Use of PREDICT Electronic
Decision Support Within Primary Care

• RNZCGP Conference 2007

Dr Janine Bycroft on behalf of the research team
Dr Tania Riddell, Dr Sue Wells, A/P Tim
Kenealy, Prof Rod Jackson, University of Auckland
and Paul Roseman and Kate Moodabe, ProCare Health
Ltd
2
Overview

• Background
• Study Aim Objectives
• Study Design
• Methods
• Results
• Discussion
• Conclusion

3
1. Background

• In a population of 10,000 primary care patients,
  each year there are about
• 1 diabetic death
• 5 breast cancer registrations
• 1 cervical cancer registration
• 1 suicide
• 80-100 coronary stroke events
• For NZ,
• 9,000 deaths from CVD each year
• compared to 500 deaths on the roads, and
• 600 from breast cancer

BMJ 19943081019-22
4
Large gap between best care usual care

• 25,384 (men gt 45 yrs women gt 55 yrs)
• Electronic records of patients visiting GP
• Dunedin RNZCGP research data base
• TC, HDL-C SBP only documented in 1 out of 4
• CVD risk could be estimated for 1 in 3
• 28 with hx CVD on statins BP?Rx
• 16 with CVD risk gt15 on statins BP?

Rafter et al NZMJ 2005
5
PREDICT brings these documents to the point of
care

• Web-based decision support system for CVD and
  Diabetes
• Integrated with main PMS systems such as
  Med-tech
• Assists with CVD risk assessment and management
  
• Provides individualised recommendations for
  management based on latest clinical guidelines

6
PREDICT-CVD Evaluation Study 2004 (NZMJ 2006)
four-fold increase in risk assessment
7
Persisting Questions

• Who tends to use PREDICT the most?
• Are there any significant differences between
  user groups?
• What are the most common barriers challenges
  to using PREDICT?
• What factors enable and support practices to
  start or use PREDICT more?
• Popular images of power users and non-users
  true or myths?

8
2. Study Aim Objectives

• Aim
• to describe the patterns of use, barriers,
  challenges and enablers to the use of PREDICT by
  New Zealand primary healthcare providers
• Specific objectives
• Investigate and describe the use of PREDICT by
  healthcare practitioners in ProCare primary care
  practices in Auckland
• Identify characteristics of frequent users
  compared to infrequent users of PREDICT
• Describe barriers, enablers and motivators for
  PREDICT use
• Identify changes that will lead to increased use
  of PREDICT within primary care

9
3. Study Design

• Stage 1 Qualitative interviews - key informant
  interviews of PHO staff focus groups (GPs
  nurses)
• Stage 2 Development of a questionnaire informed
  by key themes identified in Stage 1
• Distribution to approx 500 GPs/nurses
• Data collection, entry and subsequent qualitative
  and quantitative analyses
• Stage 3 Data collection Database matching to
  provide descriptive quantitative data analysis
  (any patterns of use, use over time, practice
  characteristics etc )

10
4. Methods

• Setting Primary care greater Auckland region
• Population ProCare general practice teams
• Samples
• 7 key informant interviews
• 5 Focus groups with a total of 42 participants
  (16 GPs 26 practice nurses)
• Questionnaire to 489 GPs practice nurses
• Response rates of 84.5 for the GPs (n262)
• 68 for nurses
  (n122)
• Data Analyses
• Qualitative NVivo 7
• Quantitative - SAS STATA

11
5. Results Only time to report some findings
from questionnaire

• Main users of PREDICT are GPs
• Since 2002, 450 GPs users and 90 nurses
• 2. PREDICT Users were more likely to
• Work between 5 9 sessions/week
• Work in a larger practice
• ProCare Network Auckland gt PNN and PNM
• Have vocational GP registration
• 3. Remaining variables not statistically
  significant
• Age, gender, number patients seen per wk,
• average hours/wk, number nurses,
• practice funding formula, location, high needs.
  

12
Internet use within a consultation

• At least once a week
• 53.4 of GPs and 28.8 of nurses
• At least once a month
• 72.1 of GPs and 43.8 of nurses at least once
  per month.
• Rarely (lt1/month) or never
• 20 for GPs and 40.5 for nurses
• Recommend a website for patients to look up
• At least once a week
• 43.4 GPs, 24.8 nurses
• At least once a month
• Another 44.6 GPs and 53.7 nurses
• Less than once a month
• 20.2 GPs and 22.3 nurses
• Rarely or never
• - 12 GPs and 21.5 nurses

(GPs n262, Practice nurses n122)
13
Screening prevention priorities within primary
care
14
What do GPs nurses rank as their top 3
population preventive/screening activities?

• For GPs, the most common activities to be ranked
  within the top three were
• 1st - cervical screening, at 68
• 2nd equal - Mammography 62
• 2nd equal - Cardiovascular risk assessment at 62
• 4th childhood vaccinations at 48
• For nurses,
• 1st - cervical screening, at 87
• 2nd - childhood vaccinations at 77
• 3rd - mammography at 48
• 4th cardiovascular risk assessment at 39

(If limited time, money and resources)
15
CVD Risk Assessments who how

• Who predominantly does CVDRA in your practice?
• 58-70 time - GPs
• 29-37 - GPs and nurses
• 0.5 2.5 Nurses when they can
• 0.4 2.5 Nurses with protected time for nurse
  clinics
• How?
• Most CVD risk assessments are done
  opportunistically
• 78.8 of GPs and 64.7 of nurses
• Only small proportion of health providers do CVD
  risk assessments systematically
• 11.2 of GPs and 16.8 of nurses reported
  systematic CVDRA in their practice using recall
  or reminder systems

16
Preferred method for CVD risk assessment

• For GPs
• 20 - informally by looking at their risk
  factors
• 24.2 - paper-based risk charts
• 11.5 - risk calculator within their PMS
• 36.5 - prefer to use PREDICT (PROMPT or
  CVD/Diabetes)
• 7.7 - prefer to use a combination of the above
  options
• For practice nurses
• 40 - informally by looking at their risk
  factors
• 8.7 - using the paper-based risk charts
• 14.1 - risk calculator within their PMS
• 30.4 - prefer to use PREDICT (PROMPT or
  CVD/Diabetes)
• 6.5 - prefer to use a combination of the above
  options

Poor accuracy
17
What GPs are saying

• I use version 2 consistently and I mainly use it
  as an aid to help me decide, in the same way x
  was, in terms of motivation of patients to make
  lifestyle changes and also to make the right
  decision about whether or not the person whos
  got moderate cholesterol should be on a statin.
  (Male GP)
• I use it frequently and I find that the patients
  respond well they like seeing you put it into a
  computer programme. And they like seeing an
  analysis. And I print off very often the sheet
  to give to them, which really lays out the
  lifestyle changes. I also find it helpful to
  assist me in making decisions about statins Im
  sometimes surprised when I would have thought
  that I should put a patient on a statin, the
  Predict might come up with only a 10 risk and
  suggest not. (Female GP)

18
Barriers

• Most common barriers appeared to be
• cost - patient wont pay
• limited resources (space and nurse time)
• fit with workflow
• time
• concern patients would be reluctant
• roles not defined
• lower priority within practice activities
• Sometimes
• Technological computer old slow, mapping
  issues, PREDICT not yet integrated with all PMSs
• Rarely important
• patient confidentiality or access to patient data
• resistance by practice leaders or management
• Usability

19
Feedback from nurses

• Doctors not willing
• GPs prefer to do it themselves
• Havent used it, could do with some education
  and time to implement if GPs agree to us nurses
  using it
• GPs only do PREDICT. Time and space are issues
  for nurses.

20
Additional focus group comments

• I think the concept of doing it through the
  nurse, I mean Ive never thought of that, but I
  think its a fantastic concept. I mean to have
  that extra information on, if you just think of
  it from a population risk point of view in the
  next 40 or 50 years, that informations going to
  prove absolutely valuable.
• And even if you dont have the time to get a
  nurse, if I have to get one of my nurses back on
  a Monday.
• (Male GP)

21
6. Discussion

• Myth Busters
• Image of Power users middle age male GPs
• Not used by older GPs or nurses
• Nurses not interested or able to do CVDRA

BUSTED
22
Benefits of systematic CVDRA

• Improved clinical records (higher rates of
  recording FHx, smoking, wt, ht, BP, chol, CVD
  Risk, etc)
• Increased rates of CVDRA within the community
• Improved job satisfaction (esp. for nurses)
• Increased practice income
• With time, can become a social norm

Likely to improve patient outcomes and
opportunity to prevent premature cardiac events
in some of your patients
23
Enablers to increase use of PREDICT

• CVDRA prioritised by practice staff
• GPs delegated CVDRA to nurses
• Nurses given protected time to do clinics for
  CVDRA, chronic care etc
• Clinical champion (GP or nurse, ideally both)
• Efficient practice systems processes for small,
  incremental changes

24
7. Conclusion

• Use of PREDICT
• Improves clinical records rates of CVDRA
• Considered easy to use and learn
• Robust validation evaluation, ongoing research
  and dev.
• Users vs non-users
• More likely to be in a larger practice,
• Highest rates of CVDRA
• occurring in practices where predominantly done
  by nurses with dedicated time

25
Conclusion continued

• CVD is the leading cause of premature death
  morbidity in NZ
• Yet largely preventable
• Estimated 80 of premature heart disease, stroke
  and diabetes could potentially be prevented!
• Good treatment and management is available
• To address this issue,
• GPs and practice nurses are critical
• We need to ensure primary care is prioritizing
  CVD,
• move towards systematic CVDRA
• Tools to assist
• This study confirms GPs and practice nurses find
  PREDICT is a useful tool for risk assessment
  management

26
Acknowledgements

• All the GPs and practice nurses for their
  contribution and time
• PREDICT Team, Epi Biostatistics, University of
  Auckland
• Especially Drs Tania Riddell, Sue Wells, Tim
  Kenealy, Prof Rod Jackson,
• Analysts -Shaheen Sultana, Mildred Lee, Joanna
  Broad
• ProCare Health staff who assisted in many ways