Table 4.
Number and Costs of Health Care Resources for Each Cohorta
| Unadjusted Costsb | ||||||
|---|---|---|---|---|---|---|
| Variable | Abusers (n = 77,369) | Controls (n = 77,369) | P Valuec | At Risk (n = 150,214) | Controls (n = 150,214) | P Valuec |
| Inpatients | ||||||
| Pre-index visits | 0.2 (0.6) | 0.1 (0.3) | <0.0001 | 0.1 (0.4) | 0.1 (0.3) | < 0.0001 |
| Post-index visits | 0.3 (0.8) | 0.1 (0.3) | <0.0001 | 0.2 (0.6) | 0.1 (0.3) | < 0.0001 |
| Pre-index costs | 7,042 (36,035) | 2,686 (23,648) | <0.0001 | 5,399 (35,763) | 2,769 (21,807) | < 0.0001 |
| Post-index costs | 12,653 (56,724) | 2,701 (22,759) | <0.0001 | 9,940 (46,119) | 3,151 (21,819) | < 0.0001 |
| Annual costs | 19,695 (72,146) | 5,387 (33,899) | <0.0001 | 15,338 (61,922) | 5,920 (32,111) | < 0.0001 |
| Outpatients | ||||||
| Pre-index visits | 2.8 (4.9) | 1.6 (3.5) | < 0.0001 | 2.4 (4.5) | 1.9 (3.8) | < 0.0001 |
| Post-index visits | 3.0 (5.2) | 1.7 (3.7) | < 0.0001 | 2.9 (5.1) | 1.9 (3.8) | < 0.0001 |
| Pre-index costs | 6,557 (29,744) | 5,387 (32,749) | < 0.0001 | 6,055 (29,122) | 5,634 (31,217) | < 0.0001 |
| Post-index costs | 6,892 (30,797) | 6,365 (37,987) | < 0.0001 | 7,311 (33,033) | 6,000 (32,720) | < 0.0001 |
| Annual costs | 13,449 (55,962) | 11,752 (67,850) | < 0.0001 | 13,366 (57,526) | 11,634 (61,491) | < 0.0001 |
| Emergency | ||||||
| Pre-index visits | 3.4 (7.7) | 0.6 (1.6) | < 0.0001 | 1.1 (2.4) | 0.5 (1.4) | < 0.0001 |
| Post-index visits | 3.9 (7.9) | 0.6 (1.6) | < 0.0001 | 1.7 (3.0) | 0.5 (1.4) | < 0.0001 |
| Pre-index costs | 2,494 (5,634) | 460 (1,372) | < 0.0001 | 805 (1,943) | 389 (1,178) | < 0.0001 |
| Post-index costs | 3,518 (6,496) | 495 (1,410) | < 0.0001 | 1,593 (2,896) | 431 (1,269) | < 0.0001 |
| Annual costs | 6,011 (11,150) | 955 (2,296) | < 0.0001 | 2,398 (4,119) | 820 (1,991) | < 0.0001 |
| Pharmacy | ||||||
| Pre-index costs | 3,520 (4,984) | 1,895 (3,988) | < 0.0001 | 2,520 (3,910) | 1,640 (3,107) | < 0.0001 |
| Post-index costs | 3,518 (5,289) | 1,975 (4,352) | < 0.0001 | 2,601 (3,956) | 1,672 (3,322) | < 0.0001 |
| Annual costs | 7,038 (9,645) | 3,870 (8,086) | < 0.0001 | 5,121 (7,468) | 3,312 (6,230) | < 0.0001 |
| Total cost | ||||||
| Pre-index costs | 19,614 (50,681) | 10,428 (42,599) | < 0.0001 | 14,779 (49,214) | 10,432 (40,314) | < 0.0001 |
| Post-index costs | 26,580 (68,639) | 11,536 (46,794) | < 0.0001 | 21,445 (60,045) | 11,253 (41,581) | < 0.0001 |
| Annual costs | 46,194 (100,608) | 21,964 (80,531) | < 0.0001 | 36,224 (91,735) | 21,685 (74,004) | < 0.0001 |
| Adjusted Post-Index Costsd | ||||||
| Inpatient | 12,600 (11,220) | 2,670 (3,251) | < 0.0001 | 9,861 (7,922) | 3,254 (3,357) | < 0.0001 |
| Outpatient | 7,580 (26,042) | 6,223 (24,559) | < 0.0001 | 7,834 (26,683) | 6,335 (25,254) | < 0.0001 |
| Emergency | 3,476 (2,167) | 509 (397) | < 0.0001 | 1,589 (753.3) | 435 (313) | < 0.0001 |
| Pharmacy | 3,667 (2,659) | 1,914 (1,085) | < 0.0001 | 2,661 (1,529) | 1,642 (944) | < 0.0001 |
| Total | 33,942 (54,445) | 10,754 (17,613) | < 0.0001 | 24,208 (36,164) | 11,692 (21,806) | < 0.0001 |
aValues in this table are shown as mean (SD).
bIncludes patients with zero costs.
cKruskal Wallis test P values for overall abuse versus overall control.
dMedical costs were adjusted for differences in patient demographics and comorbidities between abuse/dependence patients and their matched controls by claim type, subtotal, and total, using multivariate regression analyses. Because cost data are not normally distributed, a 2-step model was used. In the first step, a logit regression model predicted a probability for each patient of having costs by claim type, as well as for subtotal and total costs based on patient age, gender, race, region, poisonings, and comorbidities. The second step used a log normal regression analysis to estimate conditional costs for patients with nonzero costs and also adjusted for age, gender, race, region, poisonings, and comorbidities. This adjusted log transformed cost was retransformed then multiplied by the patient’s probability of having a nonzero cost. This step adjusted the estimated mean cost downward to account for zero cost patients.
SD = standard deviation.