A total of 3,110 records were identified for potential inclusion. After exclusion of 1,181 records (Fig. 1) with ineligible CPT/DRG codes (i.e., not indicating colorectal surgery, n = 551), ineligible ICD codes (i.e., not indicating cancer or non-cancer indication, n = 19), invalid length of stay (i.e., negative, n = 13), Medicare/Medicaid dual eligibility (n = 532), repeat surgery between 90–180 days after the initial surgery (n = 48), admission date outside the study period (n = 3), or buprenorphine prescriptions within the 1 year prior to admission (n = 15), 1,929 patients were included.
The cancer cohort consisted of 891 patients, and 1,038 patients were included in the non-cancer cohort. Patients in the cancer cohort were older (median, interquartile range [IQR] age 55 [49–60] years vs. 48 [39–56] years, p < 0.001) and were more likely to be male (53.8% vs. 47.5%, p = 0.006). Patients in the cancer cohort were more likely to live in a rural county (55.8% vs. 46.1%, p < 0.001) and had a longer travel distance (20.1 [7.1–52.3] miles vs. 14.3 [5.8–38.0] miles, p < 0.001). Cancer patients were less likely to have used opioids in the year prior to surgery (42.0% vs. 48.9%, p = 0.002), but other preoperative opioid use characteristics were similar. Patients with cancer had shorter hospital length of stay (5 [3–9] days vs. 6 [3–10] days, p < 0.001), were less likely to receive a postoperative opioid prescription (47.1% vs. 52.7%, p = 0.015) and more likely to receive a postoperative acetaminophen prescription (1.6% vs. 0.6%, p = 0.032). Patient characteristics are presented in Table 1.
Table 1
Patient demographics, procedures, and medications for Kentucky Medicaid beneficiaries undergoing surgery with and without cancer, 2014–2018
|
Cancer
(n = 891)
|
Non-Cancer (n = 1,038)
|
p value
|
Demographic Characteristics
|
Age, years
|
55 [49–60]
|
48 [39–56]
|
< 0.001
|
Sex, n (%)
|
|
|
0.006
|
Female
|
412 (46.2)
|
545 (52.5)
|
|
Male
|
479 (53.8)
|
493 (47.5)
|
|
Race, n (%)
|
|
|
0.140
|
Black
|
70 (7.9)
|
79 (7.6)
|
|
White
|
603 (67.7)
|
743 (71.6)
|
|
Other/Not Reported
|
218 (24.5)
|
216 (20.8)
|
|
Residential RUCC, n (%)
|
|
|
< 0.001
|
Metropolitan (1–3)
|
394 (44.2)
|
560 (53.9)
|
|
Rural (4–9)
|
497 (55.8)
|
478 (46.1)
|
|
Drive Distance
|
20.1 [7.1–52.3]
|
14.3 [5.8–38.0]
|
< 0.001
|
Encounter Year, n (%)
|
|
|
0.454
|
2014–2015
|
438 (49.2)
|
528 (50.9)
|
|
2016–2018
|
453 (50.8)
|
510 (49.1)
|
|
Prehospital Opioid Use, n (%)
|
|
|
|
Past 30 Days (any)
|
249 (27.9)
|
273 (26.3)
|
0.417
|
Past 365 Days (any)
|
374 (42.0)
|
508 (48.9)
|
0.002
|
Multiple Opioid Prescriptions
|
187 (21.0)
|
214 (20.6)
|
0.841
|
Charlson Comorbidity Index
|
2 [1–3]
|
1 [0–2]
|
< 0.001
|
Expansion Program
|
561 (63.0)
|
652 (62.8)
|
0.946
|
Procedural Characteristics
|
Procedure Class, n (%)
|
|
|
0.150
|
Major
|
596 (66.9)
|
726 (69.9)
|
|
Minor
|
295 (33.1)
|
312 (30.1)
|
|
Procedure, n (%)*
|
|
|
0.303†
|
Laparoscopic
|
295 (35.7)
|
312 (33.3)
|
|
Partial Colectomy
|
166 (20.1)
|
163 (17.4)
|
0.153‡
|
Colectomy
|
89 (10.8)
|
126 (13.5)
|
0.084‡
|
Other (Proctectomy, Proctocolectomy)
|
40 (4.8)
|
23 (2.5)
|
0.007‡
|
Open
|
532 (64.3)
|
624 (66.7)
|
|
Partial Colectomy
|
375 (45.3)
|
481 (51.4)
|
0.011‡
|
Colectomy
|
82 (9.9)
|
129 (13.8)
|
0.013‡
|
Other (Proctectomy, Proctocolectomy)
|
75 (9.1)
|
14 (1.5)
|
< 0.001‡
|
Hospital Length of Stay, days
|
5 [3–9]
|
6 [3–10]
|
< 0.001
|
Medications
|
|
|
|
Chemotherapy, n (%)
|
|
|
|
Adjuvant
|
146 (16.4)
|
-
|
-
|
Neo-Adjuvant
|
50 (5.6)
|
-
|
-
|
Both
|
46 (5.2)
|
-
|
-
|
Discharge Analgesics, n (%)
|
|
|
|
Opioid Analgesic
|
420 (47.1)
|
547 (52.7)
|
0.015
|
MME per day
|
50.0 [36.0-67.5]
|
50.0 [37.5–70.3]
|
0.871
|
Days' supply
|
5 [4–8]
|
5 [4–8]
|
0.230
|
Acetaminophen
|
14 (1.6)
|
6 (0.6)
|
0.032
|
Gabapentinoid
|
33 (3.7)
|
53 (5.1)
|
0.137
|
NSAID
|
38 (4.3)
|
38 (3.7)
|
0.497
|
Skeletal Muscle Relaxant
|
26 (2.9)
|
33 (3.2)
|
0.740
|
Values presented as n (%) when categorical or median [interquartile range] when continuous.
RUCC, rural urban continuum code classification; NSAID, nonsteroidal anti-inflammatory drug
*n = 1,763 (827 cancer, 936 non-cancer) after excluding 166 patients without specific current procedural terminology codes that were included in the larger cohort based on diagnosis related group.
† p value for Chi-square test comparing laparoscopic vs. open procedures
‡ p value for Fisher’s exact test comparing specific procedure (e.g., laparoscopic partial colectomy) vs. all other procedures
Using unadjusted analysis, patients in the cancer cohort were more likely to develop long-term opioid use (40.9% vs. 34.8%, p = 0.006). These patients had more total days of opioid therapy through 180 days after discharge (10.0 [0.0-75.5] days vs. 7.0 [0.0–30.0] days, p < 0.001).
Multiple logistic regression results are presented as odds ratio plots in Fig. 2. Adjusting for other variables, patients with cancer were significantly more likely to develop persistent opioid use (OR 1.35, 95% CI 1.08–1.69). Other variables independently associated with an increased likelihood of persistent opioid use included opioid use within the 30 days prior to admission (OR 3.00, 95% CI 2.34–3.85), travel distance > 60 miles (OR 1.72, 95% CI 1.26–2.32), opioid use between 12 months and 1 month prior to admission (OR 1.71, 95% CI 1.35–2.16), travel distance between 30 and 60 miles (OR 1.45, 95% CI 1.06-2.00), and male sex (OR 1.25, 95% CI 1.02–1.54). Additionally, compared to patients undergoing laparoscopic colectomy, patients were more likely to develop persistent opioid use if they underwent laparoscopic proctocolectomy (OR 2.66, 95% CI 1.19–5.91) or open colectomy (OR 1.75, 95% CI 1.37–2.23). Factors associated with a lower likelihood of developing persistent opioid use were age > 65 (OR 0.12, 95% CI 0.02–0.64), fee-for-service as primary payer (OR 0.38, 95% CI 0.22–0.67), race other than Black or White (OR 0.67, 95% CI 0.51–0.87), and encounter year in 2016–2018 (OR 0.77, 95% CI 0.62–0.97).
Several factors were significant in the unadjusted regression model but did not remain significant after adjustment. These included CCI, filling a discharge opioid prescription, increasing daily MME and days’ supply of the first opioid prescription, and filling a gabapentinoid or skeletal muscle relaxant prescription (see supplemental Fig. 1).