Table 4.
Adjusted Health Care Resource Utilization and Costs During 1-Year Follow-up Among the 3 Treatment Pattern Groupsa
| Adjusted All-Cause Outcomes | Switchers vs. Nonswitchers | Discontinuers vs. Nonswitchers |
|---|---|---|
| Health care resource utilization (count outcomes), IRR (95% CI) | ||
| Hospitalizationsb | 1.13 (0.84-1.53) | 2.05 (1.70-2.48) |
| ED visitsb | 1.10 (0.93-1.30) | 1.50 (1.34-1.68) |
| Outpatient visitsc | 1.18 (1.16-1.21) | 1.01 (1.00-1.03) |
| Health care costs ($), mean difference (95% CI)d | ||
| Total all-cause health care costs | 10,120 (8,235-12,033) | –18,611 (-$20,254 to -$17,025) |
| Medical costs | 2,746 (1,090-4,677) | 3,729 (1,970-5,527) |
| ED costs | 170 (15-341) | 298 (198-409) |
| Inpatient costs | 482 (29-1,470) | 2,578 (2,445-2,973) |
| Outpatient costs | 2,028 (1,470-3,067) | 525 (29-1,394) |
| Prescription drug costs | 8,988 (7,535-10,610) | –20,486 (-$21,319 to -$19,636) |
aCovariates included age, sex, geographic region, insurance plan type, index drug, index year, baseline Charlson Comorbidity Index, baseline health care resource utilization, baseline total health care costs, and baseline nonbiologic medications.
bHospitalizations and ED visit IRRs were derived from a negative binomial model.
cOutpatient visit IRRs were derived from a Poisson model.
dHealth care costs were modeled using generalized linear regression with a log link and gamma distribution or 2-part models. For 2-part models, logistic regression was used to model probabilities of positive costs followed by gamma regression to model costs among patients with positive costs.
CI = confidence interval; ED = emergency department; IRR = incidence rate ratio.