Table 2.

Adjusted odds ratio of nonsteroidal anti-inflammatory drug start (or stay on) with eGFR, proteinuria, and nephrology visit (visit-pair n=15,917)

CovariateaStay off n (Row %)Start n (Row %)Adjusted OR (95% CI) of Starting NSAIDStop n (Row %)Stay on n (Row %)Adjusted OR (95% CI) of Staying on NSAID
N visit-pairs (row %)n=14,075 (88.3)n=1842 (11.7)
Within-stratum n visit-pairs (row %)13,377 (95.0)698 (5.0)857 (46.5)985 (53.5)
eGFR at antecedent visit, ml/min per 1.73 m2Mean 41.6±13.9Mean 49.6±13.81.44 (1.34 to 1.56)bMean 48.0±14.151.0±12.61.06 (0.95 to 1.18)b
Urine protein at antecedent visit, g/24 h
 <0.13526 (93)254 (7)cReference317 (43423 (57)Reference
 0.1 to <0.53181 (95)168 (5)0.91 (0.73 to 1.13)204 (47)227 (53)1.00 (0.89 to 1.49)
 ≥0.53112 (97)112 (3)0.82 (0.62 to 1.08)138 (62)86 (38)0.77 (0.53 to 1.11)
Seeing a nephrologist at antecedent visit
 No2660 (92)247 (8)Reference320 (38)523 (62)Reference
 Yes10,717 (96)451 (4)0.70 (0.56 to 0.87)537 (54)462 (46)0.61 (0.46 to 0.81)
Pain at antecedent visit interfering moderately or greater with normal work
 No9171 (96)424 (4)Reference495 (47)559 (53)Reference
 Yes4038 (94)266 (6)1.48 (1.19 to 1.85)352 (46)410 (54)1.15 (0.89 to 1.49)
  • OR, odds ratio; 95% CI, 95% confidence interval; NSAID, nonsteroidal anti-inflammatory drug.

  • a The transition model included the following variables: NSAID use at antecedent visit, black race, sex, age group, eGFR, urine protein group, body mass index group, diabetes, cardiovascular disease, hypertension, arthritis, seeing a nephrologist, pain interfering with normal work moderately or greater (all at antecedent visit except for black race and sex, which were evaluated at baseline), years from baseline visit to the subsequent visit, and the interactions of these variables with NSAID use at antecedent visit.

  • b OR comparing participants with a 10 ml/min per 1.73 m2 higher eGFR level at antecedent visit with those with a specific eGFR level.

  • c The row percentage of NSAID switch at each urine protein stratum or nephrologist visit stratum is conditioned on NSAID use or not at the antecedent visit: for example, the row percentage of No to Yes = (number of No to Yes pairs)/[(number of No to No pairs) + (number of No to Yes pairs)].