TABLE 4

Relative Difference in Change Over Time in Health Expenditures Between SDM Groupsa

Total Health Expendituresb
(95% CI)P ValuecOut-of-Pocket Expendituresb
(95% CI)P ValuecOffice-based Expendituresb
(95% CI)P ValuecInpatient Expendituresb
(95% CI)P ValuecEmergency Department Expendituresb
(95% CI)P ValuecPrescription Expendituresb
(95% CI)P Valuec
SDM increasing
vs unchanged high−260 (–620, 99).2−53 (–149, 44).3−67 (–157, 24).1−61 (–152, 30).2−28 (–52, –3).0312 (–68, 93).8
 vs unchanged low−320 (–850, 210).2−6 (–123, 110).910 (–129, 150).9−112 (–243, 19).09−16 (–46, 14).3−11 (–107, 85).8
 vs decreasing−584 (–1131, –38).04−142 (–265, –19).02−107 (–245, 30).1−1 (–128, 126).9−31 (–65, 3).0728 (–69, 125).6
SDM unchanged high
vs unchanged low60 (–427, 546).8−46 (–137, 44).3−77 (–196, 42).251 (–64, 166).1−12 (–37, 2).423 (–58, 105).6
 vs decreasing−324 (–790, 142).2−89 (–195, 16).1−41 (–171, 89).560 (–44, 164).3−3 (–36, 30).815 (–75, 106).7
SDM unchanged low
vs decreasing−264 (–890, 362).4−135 (–248, –23).02−118 (–286, 50).2110 (–27, 248).1−15 (–52, 22).439 (–63, 140).4
  • a Because the inclusion of covariates did not change the association of SDM with expenditures, unadjusted results are shown.

  • b The highest 2.5% of expenditure values were trimmed at the 97.5 percentile to avoid bias and ensure that assumptions of normality of the statistical models were not violated. All amounts are in US dollars ($).

  • c P values calculated from linear regression with robust variance estimation accounting for the complex survey design.