Title: Quality of Life in Breast Cancer Survivors: A Latent Growth Analysis
1Quality of Life in Breast Cancer Survivors A
Latent Growth Analysis
- Lu Jing
- MS, HPA
- Joe Vasey, PhD
- Research Associate, Center for
- Health Care and Policy Research
2Breast Cancer in the US
- Breast cancer is the most common type of cancer
among women - Breast cancer accounts for 31 of new cancer
cases among women nearly 204,000 cases in 2002
(ACS, 2002) - Other common cancers among women include
- Lung cancer (12)
- Colo-rectal cancer (12)
- Uterine cancer (6)
- Ovarian cancer (4)
- 13 of all US women will eventually develop
breast cancer - Breast cancer is survivable, and the probability
of surviving is improving (ACS, 2002) - 1974 5-year survival rate was 75
- 1992 5-year survival rate was 86
- 5-year survival is strongly related to stage at
diagnosis - Local (Stage I) 96
- Regional (Stage II) 78
- Distant (Stage III) 21
3Quality of Life
- The increased probability of surviving cancer has
led to greater interest in survivors lives after
diagnosis and treatment. - After treatment survivors may experience a
variety of physical and psychological problems - Quality of Life (QoL) is concerned with The
complete social and psychological being the
individuals performance of social roles, her
mental acuity, her emotional state, her sense of
well being, and her relationship with others.
(Levine, 1987) - Multidimensional .
- Physical well-being the control, coping, or
relief of physical symptoms and the maintenance
of function and independence - Psycho-social - Maintenance of a sense of
control, cognitive and emotional functioning,
roles, and relationships with others in a variety
of settings - Spiritual the maintenance of a sense of hope,
meaningfulness, and purpose to life
4Quality of Life Research
- Cross-sectional studies at a single point in
time, what associations exist between QoL
measures and various surivivor characteristics
(demographic, medical, etc)? - Longitudinal studies over a period of time,
what changes occur in QoL measures, and what
accounts for these changes?
5Summary of FindingsQoL Level
6Summary of FindingsChange in QoL
- Poorly studied 5 studies
- All rely on mean QoL levels across time
- Conflicting findings
- Maximum physical and psychological QoL achieved
1-2 years after treatment. QoL measures declined
in 2nd and 3rd years - Highest physical and psychological QoL occurred
in 2nd through 5th year after treatment. Lower
levels were observed immediately after treatment
and long term (5 years)
7Purpose
- Address shortfalls in existing research
- Describe and model individual trajectories in
QoL, rather than mean differences across time - Identify predictors of change in QoL
- Address limitations of traditional analytic
methods - Recognize that observed scores are not true
scores (measurement error) - Account for varying temporal points of
observation across participants - Account for time-varying covariates
- Maximize the use of available data
8Method
- Sample
- Part of a larger observational study of 1763
cancer survivors - 25-62 years of age at time of diagnosis
- Stage I, II, or III
- 1st diagnosis between January 1997 and December
1999. - Data collection began in the fall of 2000, so at
the time of their first interview survivors were
between 1 to nearly 5 years post-diagnosis.
9Timing of Data Collection
Time frame covered by four annual interviews was
determined by time since diagnosis at the time of
the first interview
W1 W2 W3 W4
W1 W2 W3 W4
W1 W2 W3 W4
W1 W2 W3 W4
1 2 3 4
5 6 7 8
Years since diagnosis
10Sample Characteristics
- 405 out of 550 breast cancer survivors completed
all four interviews - Average age at diagnosis 52.4
- 15 under 45, 50 45-55, 35 over 55
- Average time between diagnosis and 1st interview
2.7 years - Race Predominantly white (94)
- Education
- 2 less than HS, 31 HS, 25 some college, 20
college, 22 post college - Cancer stage at diagnosis
- 50 Stage I, 44 Stage II, 6 Stage III
- Recurrance of cancer 20
- Treatment Status at 1st interview
- 77 not in treatment, 7 in treatment for active
cancer, 16 in treatment but no active cancer - Other conditions
- 61 reported 1 or more chronic conditions
- Marital Status
- 84 married or living with a partner
- Employment
- 59 were working at all four interviews
- 25 were working intermittently
11Measures
- Quality of Life
- SF-12 Physical and Mental Scores
- Demographic
- Age, race, education, marital status, employment
status, home ownership - Medical
- Stage at diagnosis, recurrance, treatment status,
presence of chronic health conditions - Time
- Time since diagnosis
12Analysis Plan
- Latent growth modeling
- Unconditional means model does variation in QoL
measures exist, and is it sufficient to warrant
further analysis? - Unconditional latent growth model to what extent
does time account for significant variability in
QoL among participants? - Conditional latent growth model to what extent
do demographic and medical/health covariates
account for significant variability?
13Unconditional Means Model
- Significant within-person variance and
between-person variance for both physical and
mental QoL measures - Within Individuals QoL changed over time
- Between Individuals differed from one another
- Significant variability exists, some of which may
be may be explainable by time from diagnosis.
14Individual SF-12 Trajectories
15Change in Mean SF-12 Scores
16Unconditional Latent Growth Model
plt.1, plt.05, plt.001
- Significant within person variance in both
physical and mental QoL suggest that - survivors deviate significantly from their
predicted trajectories - Significant variability exists for both initial
status and rate of change for both - measures
- substantive covariates are needed to explain
further individual differences.
17?0 initial status (intercept) D1 variance in
initial status ?1 rate of change (slope) D2
variance in rate of change Yi QoL at time I Ti
individually varying point in time
Covariates
18Latent Growth Model with Covariates
plt.1, plt.05, plt.001
- Physical QoL negative but non-significant rate
of change - Mental QoL positive but non-significant rate of
change - both measures Variance in initial status and
rate of change declined compared to - the unconditional model, indicating covariates
have some explanatory power - negative covariance survivors with higher
levels of QoL 1 year after diagnosis - experienced lower rate of change in QoL over
time
19ML Estimates For Effects of Covariates Physical
QoL
plt.1, plt.05, plt.001
20Summary Physical QoL
Negative significant covariance suggests that
survivors with better initial QoL Experienced
smaller rates of change over time.
21ML Estimates For Effects of Covariates Mental
QoL
plt.1, plt.05, plt.001
22Summary Mental QoL
Negative significant covariance suggests that
survivors with better initial QoL experienced
smaller rates of change over time.
23Summary
- Mean (group) physical and mental QoL measures are
stable over time - Significant individual differences exist in
change trajectories over time - Diverse trajectory patterns point to the need to
better understand the process of QoL development
and change over time - Investigation of additional covariates is needed
to explain the heterogeneity in individuals
initial status and rates of change
24Limitations
- Sample homogeneity
- Race (largely white)
- Geographic location (largely rural)
- Alternative QoL measures
- Is the SF-12 too general a measure for the issues
confronting this sample? - Breast cancer specific instruments?
- Additional (currently unidentified) covariates
- Longer term changes (8 years)