Table 2.
Age at presentation and (a) overall relationship with overall perceived quality of life (good/poor)a and (b) relationship with quality of life ordered from 1 (‘best’) to 5 (‘worst’). In each case model estimates are adjusted for gender, albumin ratio, UDCA response and disease durationb
| (a) | ||||||
|---|---|---|---|---|---|---|
| Outcome | Covariate | (S.E.) | OR (CI) | Z value | P value | Pseudo R 2 |
| Good vs. Poor Quality of life | Age at presentation (10 unit increase) | −0.16 (0.07) | 0.86 (0.75–0.98) | −2.30 | 0.02 | 0.03 |
| Male | −0.04 (0.23) | 0.97 (0.61–1.51) | −0.15 | 0.88 | ||
| Albumin ratio | −0.56 (0.38) | 0.57 (0.27–1.20) | −1.46 | 0.15 | ||
| UDCA responder | −0.28 (0.16) | 0.76 (0.56–1.04) | −1.75 | 0.08 | ||
| Disease duration | 0.02 (0.01) | 1.02 (1.00–1.05) | 1.97 | 0.05 | ||
| (b) | |||||
|---|---|---|---|---|---|
| (S.E.) | OR (CI) | t value | P value | Pseudo R 2 | |
| Age at presentation (10 units) | −0.16 (0.06) | 0.85 (0.76, 0.96) | −2.75 | 0.006 | 0.04 |
| Male | −0.25 (0.20) | 0.78 (0.53, 1.15) | −1.24 | 0.21 | |
| Albumin ratio | −0.79 (0.32) | 0.45 (0.24, 0.85) | −2.48 | 0.01 | |
| UDCA responder | −0.36 (0.14) | 0.70 (0.53, 0.91) | −2.65 | 0.008 | |
| Disease duration | 0.02 (0.01) | 1.02 (1.00, 1.04) | 1.82 | 0.07 | |
| State 1–2 | −3.17 (0.52) | −6.06 | P < 0.001 | ||
| State 2–3 | −2.19 (0.52) | −4.22 | P < 0.001 | ||
| State 3–4 | −1.32 (0.52) | −2.56 | 0.01 | ||
| State 4–5 | 0.40 (0.52) | 0.77 | 0.44 | ||
In this analysis quality of life outcome is modelled on 1015 patients; 654 reporting good and 361 reporting poor quality of life.
In this analysis quality of life outcome is modelled on 1015 patients; 233 patients reporting 1 (best quality of life), 211 patients reporting 2, 210 patients reporting 3, 267 patients reporting 4 and 94 patients reporting 5 (worst). Ordinal regression allows the ordinal nature of the global quality of life outcome (scored 1 ‘best’ to 5 ‘worst’) to be retained. The underlying assumption of proportional odds was confirmed graphically and confirming no significant difference in the likelihood ratio test comparing a proportional odds model to a multinomial‐logit (nonproportional odds) model. This analysis confirms the increasing probability of ‘better’ global quality of life impairment scores with increasing age at presentation and confirms a 10‐unit increase in age at presentation to be associated with a 15% reduction in risk of poorer quality of life (OR = 0.85, 95% CI: 0.76–0.96, P < 0.01).