Alex Mitchell www.psycho-oncology.info Department of Cancer & Molecular Medicine, Leicester Royal Infirmary Department of Liaison Psychiatry, Leicester General Hospital LOROS August 2009 LOROS August 2009 Clinical Accuracy of Cancer Clinicians Ability of health professionals to identify mood disorders
37
Embed
LOROS - Clinical Ability of Cancer Clinicians to Detect Depression (Aug09)
This is a talk from 18-Aug-09 about how well do cancer clinicians (oncologists and clinical nurse specialists) detect depression and distress in clinical practice
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Alex Mitchell www.psycho-oncology.info
Department of Cancer & Molecular Medicine, Leicester Royal Infirmary
Department of Liaison Psychiatry, Leicester General Hospital
LOROS August 2009LOROS August 2009
Clinical Accuracy of Cancer CliniciansAbility of health professionals to identify mood disorders
Clinical Accuracy of Cancer CliniciansAbility of health professionals to identify mood disorders
1. Background1. Background
What methods are used to detect mood disorders?
How often do clinicians look for mood complications?
Methods to Evaluate Depression
Conventional Scales
Short (5-10) Long (10+)
Comment: This is a reminder of the structure of the HADS scale, this version adapter for cancer.
Methods to Evaluate Depression
Conventional Scales
Ultra-Short (<5)Short (5-10) Long (10+)
Methods to Evaluate Depression
Unassisted Clinician Conventional Scales
Ultra-Short (<5) Short (5-10) Long (10+)Untrained Trained
Routine Implementation
Acceptability ?
Accuracy? Accuracy?
vsComment: schematic overview of methods to evaluate depression
n=226Comment: Frequency of cancer specialists enquiry about depression/distress from Mitchell et al (2008)
1,2 or 3 Simple QQ15%
Clinical Skills Alone73%
ICD10/DSMIV0%
Short QQ3%
Other/Uncertain9% Other/Uncertain
2%
Use a QQ15%
ICD10/DSMIV13%
Clinical Skills Alone55%
1,2 or 3 Simple QQ15%
Cancer StaffCurrent Method (n=226)
Psychiatrists
Comment: Current preferred method of eliciting symptoms of distress/depression
1,2 or 3 Simple QQ24%
Clinical Skills Alone20%
ICD10/DSMIV24%
Short QQ24%
Long QQ8%
Algorithm26%
Short QQ23%
ICD10/DSMIV0%
Clinical Skills Alone17%
1,2 or 3 Simple QQ34%
Cancer StaffIdeal Method (n=226)
Psychiatrists
Effective?
Comment: “Ideal” method of eliciting symptoms of distress/depression according to clinician
2. Primary Care - Meta-Analysis2. Primary Care - Meta-Analysis
How well do GPs (PCPs) identify depression? (clinical sensitivity)
How well do GPs (PCPs) identify the non-depressed? (clinical specificity)
How important is severity of depression/distress?
Summary
50 371 patients
9 countries
N= 108 studies
N= 41 depression studies
N= 19 depression with specificity
Predictors Examined
Severity
Age
Prevalence
Type of assessment
Duration of assessment
Comment: HSROC Curve plot for all depression detection studies from primary care
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Pre-test Probability
Pos
t-tes
t Pro
babi
lity
Baseline Probability
Depression+
Depression-
Comment: Slide illustrates Bayesian curve – pre-test post test probability for every possible prevalence
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Pre-test Probability
Pos
t-tes
t Pro
babi
lity
Baseline Probability
Depression+
Depression-
PPV
NPV
Comment: At a prevalence of 20% GPs PPV is 40% and NPV 86%
Depression vs DistressDepression vs Distress
Comment: Slide illustrates two HsROC curves, one for depression and one for distress, both from primary care. The following bayesian graph compares the two more clearly=>
0.00
0.10
0.20
0.30
0.40
0.50
0.60
0.70
0.80
0.90
1.00
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Pre-test Probability
Pos
t-tes
t Pro
babi
lity
Non-Mild Depression+
Non-Mild Depression-
Baseline Probability
Mild Depression+
Mild Depression-
Distress+
Distress-
GP Accuracy by SeverityGP Accuracy by Severity
Comment: Slide illustrates GP diagnosis of mod-severe depression is more successful than their diagnosis of “distress” or mild depression
GP Accuracy – Detection of Distress by GHQ ScoreGP Accuracy – Detection of Distress by GHQ ScoreMcCall et al (2007) Primary Care Psychiatry - Recognition by Severity
Comment: Slide illustrates raw number of people identified by severity on the GHQ. Although the % detection increases with severity, the absolute number decreased due to falling prevalence
3. Cancer Care - Meta-Analysis3. Cancer Care - Meta-Analysis
How well do cancer specialists identify depression?
How do doctors compare with nurses?
Testing Clinicians: A Meta-AnalysisTesting Clinicians: A Meta-AnalysisMethods (currently unpublished)
12 studies reported in 7 publications. 2 studies examined detection of anxiety, 8 broadly defined depression (includes HADS-T)3 strictly defined depression and 7 broadly defined distress.
9 studies involved medical staff and 2 studies nursing staff.
Gold standard tools including GHQ60, GHQ12 HADS-T, HADS-D, Zung and SCID.
The total sample size was 4786 (median 171).
Testing Clinicians: A Meta-AnalysisTesting Clinicians: A Meta-AnalysisResultsAll cancer professionalsSE =39.5% and SP =77.3%.
OncologistsSE =38.1% and SP = 78.6%; a fraction correct of 65.4%.
By comparison nursesSE = 73% and SP = 55.4%; FC = of 60.0%.
When attempting to detect anxiety oncologists managedSE = 35.7%, SP = 89.0%, FC 81.3%.
Comment: Slide illustrates Bayesian curve comparison from RCT studies of clinician with and without screening
This illustrates ACTUAL gain from screening
5. Cancer Care – Cumulative Testing5. Cancer Care – Cumulative Testing
What can enhance detection?
Cancer Population
CNS Assessment
Possible case
Depression
Screen #1+ve
n = 200 No Depression
Sp 55%
Se 70%
n = 800
N = 1000
TP = 140
FP = 360Probable Non-Case TN =440
FN = 60
PPV 28% NPV 88%
Screen #1-ve
YieldTP = 140
TN = 440
FN = 60
FP = 360
NPV 88%
PPV 28%
Sp 55%
Se 70%
Cancer Population
CNS Assessment
Possible case
Depression
Screen #1+ve
n = 200 No Depression
Sp 55%
Se 70%
n = 800
N = 1000
TP = 140
FP = 360Probable Non-Case TN =440
FN = 60
PPV 28%
Oncologist Assessment Sp 80%
Sp 40%
NPV 88%
Probable Depression TP = 56
FP = 72Probable Non-Case TN =288
FN = 84
PPV 44% NPV 77%
Screen #1-ve
Screen #2+ve
Screen #2+ve
Cumulative YieldTP = 56
TN = 728
FN = 144
FP = 72
NPV 83%
PPV 44%
Sp 91%
Se 28%
Credits & Acknowledgments
Elena Baker-Glenn University of NottinghamPaul Symonds Leicester Royal InfirmaryChris Coggan Leicester General HospitalBurt Park University of NottinghamLorraine Granger Leicester Royal InfirmaryMark Zimmerman Brown University, Rhode IslandBrett Thombs McGill University CanadaJames Coyne University of PennsylvaniaNadia Husain University of Leicester
For more information www.psycho-oncology.info
FURTHER READING:
Screening for Depression in Clinical Practice An Evidence-Based guide
ISBN 0195380193 Paperback, 416 pagesNov 2009Price: £39.99