1 / 23

Exercise and outcomes

Exercise and outcomes. Hirsch, Diehr 11-18-2005 (9 AM). Compact Results. Adjusting for Age in Regression Only Does not match Table 5, which is age-specific Big exercise effects unadjusted Much smaller differences after adjustment

ivan
Download Presentation

Exercise and outcomes

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Exercise and outcomes Hirsch, Diehr 11-18-2005 (9 AM)

  2. Compact Results • Adjusting for Age in Regression Only • Does not match Table 5, which is age-specific • Big exercise effects unadjusted • Much smaller differences after adjustment • But some quintiles have significantly fewer “years” than quintile 5 • The confidence interval doesn’t overlap with the CI for quintile 5 • Propensity score approach suggests that the covariates we used do achieve balance.

  3. Regression Model • Not in causal pathway? • gend01 age00 lage agesex lagesex white grade01 • smoke alcoh aspirin • diabada maxcom maxint maxsten copdbl shtbrt chdblmod emphysem chfbase cbd cancer prknsn06 renal arth01 • crp fib44 alb44 depscr05 cis42 benzod06 • minimental digcor • evggfp timewlk iadl adl • Possibly in causal pathway? biorea21 biores21 bmi • choladj hdl44 ldladj lipid06 • hyper aai suppul16 • Re-try propensity score model without some of these suspect variables?

  4. YOL and Exercise

  5. YHL and Exercise

  6. Years w/o ADL Difficulties

  7. Years w/o IADL Difficulties

  8. But here is the bad news • I calculated a propensity score from the variables in the regression (59 variables incl 42 in model and 17 not in) • I created quintiles of the propensity score • I looked within each quintile to see whether log kcal was correlated with the covariates.

  9. Vars in propensity score • Regression variables: gend01 age00 lage agesex lagesex white grade01 iadl adl timewlk biorea21 biores21 bmi choladj hdl44 ldladj lipid06 hyper smoke diabada alcoh aspirin maxcom maxint maxsten aai copdbl shtbrt chdblmod chfbase cbd diag01 prknsn06 renal arth01 crp fib44 alb44 depscr05 score30 digcor ih000 suppul16 cis42 benzod06 emphysem • Other Variables (obtained from screening all variables: not used in regression model, to keep sample size up): hip13 insul12 wbld23 brtsht07 domgrip benzod06 cancer arth01 aai cis42 chstpn cpvd anar1a06 avzmdia iadl hip13 insul12 married fvc18 live03 netscr03 extart07 time17 suppul16 trblsl08 ues waist13 wgt5008 white wbld23 wakeup08

  10. Exercise by propensity categories

  11. Balance: lkcal corr with: • In Propensity quintile 1, more exercise correlated with lower fibrinogen, lower wbc, lower waist circumference, lower weight (4/59 vars: a problem? Type of bias?) • 2: less hypertension • 3: higher digital symbol score, higher grip strength, married, living with spouse (vs alone, etc.) • 4: more alcohol, more aspirin, more chd, lower suppine pulse rate, lower wbc, higher grip strength • 5: higher ankle/arm ratio, lower chest pain, less trouble sleeping • 20/59*5 = 3% of all corrs sdfz • Balance is amazingly good

  12. YOL vs exercise, and vs propensity to exercise

  13. Balance if hypertension, weight, cholesterol not in model • In Propensity quintile 1, more exercise is associated with fewer IADL difficulties, lower cholesterol, lower BMI, lower LDL, lower fibrinogen, higher evggfp, smaller hips, smaller waist • 2: lower bmi, less alcohol, smaller hips, smaller waist • 3: lower bmi, higher albumin, smaller hip, more grip strength, more married, higher fvc, lives with spouse, smaller waist • 4: lower bmi, higher hdl, higher alcohol, less stenosis, higher fibrinogen, smaller hips, lower wbc, less peripherovascular disease, smaller waist • 5: No blacks, no people with 1 iadl at baseline, lower digital symbol score, less trouble falling asleep • Balance is much worse without those variables. If bmi and hip size and hdl matter, the groups are not comparable.

  14. Age-specific results Use single age group Use age-specific quintiles But Age is adjusted for in regression (I’m not sure that we want to adjust for age twice)

More Related