Association of Watson-based humanistic fast track surgery with stoma colorectal cancer recovery: A propensity matched study

Table 1 Baseline demographic and clinical characteristics of patients before and after propensity score matching, n (%)/mean ± SD

Characteristic	Pre-matched cohort					Post-matched cohort (n = 125 pairs)
	Watson-FTS (n = 165)	Routine care (n = 147)	Test statistic	P value	SMD	Watson-FTS (n = 125)	Routine care (n = 125)	Test statistic	P value	SMD
Demographics
Age, years	62.1 ± 10.8	64.5 ± 11.4	1.91	0.057	0.22	63.2 ± 10.5	63.5 ± 10.9	0.22	0.825	0.03
Male gender	96 (58.2)	85 (57.8)	0.01	0.948	0.01	72 (57.6)	71 (56.8)	0.02	0.898	0.02
BMI, kg/m2	23.6 ± 3.4	23.1 ± 3.7	1.24	0.216	0.14	23.4 ± 3.2	23.3 ± 3.5	0.24	0.814	0.03
Education (> high school)	91 (55.2)	59 (40.1)	6.99	0.008	0.31	58 (46.4)	60 (48.0)	0.06	0.799	0.03
Clinical status
ASA class III	45 (27.3)	38 (25.9)	0.08	0.776	0.03	34 (27.2)	33 (26.4)	0.02	0.887	0.02
NRS 2002 score	2.6 ± 0.9	2.5 ± 1.1	0.88	0.38	0.1	2.5 ± 0.8	2.5 ± 0.9	0	1	0
CCI score	2.4 ± 1.5	2.6 ± 1.7	1.1	0.272	0.12	2.5 ± 1.4	2.5 ± 1.5	0	1	0
Smoking (current/former)	64 (38.8)	61 (41.5)	0.24	0.624	0.06	49 (39.2)	51 (40.8)	0.07	0.796	0.03
Neoadjuvant therapy	35 (21.2)	28 (19.0)	0.24	0.627	0.05	25 (20.0)	26 (20.8)	0.02	0.879	0.02
Psychological status
HADS-anxiety	9.6 ± 3.1	7.9 ± 3.4	4.61	< 0.001	0.52	8.8 ± 2.9	8.9 ± 3.2	0.26	0.795	0.03
HADS-depression	8.9 ± 3.3	7.4 ± 3.2	4.07	< 0.001	0.46	8.2 ± 3.0	8.3 ± 2.8	0.27	0.785	0.03
Surgical/oncological
Laparoscopic approach	140 (84.8)	101 (68.7)	11.52	0.001	0.4	95 (76.0)	94 (75.2)	0.02	0.885	0.02
AJCC TNM stage III	88 (53.3)	76 (51.7)	0.08	0.775	0.03	66 (52.8)	65 (52.0)	0.02	0.898	0.02
Tumor location (rectum)	102 (61.8)	94 (63.9)	0.15	0.697	0.04	78 (62.4)	79 (63.2)	0.02	0.894	0.02
Permanent stoma	65 (39.4)	62 (42.2)	0.25	0.617	0.06	50 (40.0)	49 (39.2)	0.02	0.892	0.02
Preop albumin, g/L	36.4 ± 4.5	36.9 ± 4.8	0.95	0.343	0.11	36.6 ± 4.2	36.7 ± 4.4	0.18	0.854	0.02
Preop prealbumin, mg/L	224 ± 51	219 ± 55	0.83	0.407	0.09	221 ± 48	223 ± 50	0.32	0.748	0.04

For the prematched cohort, the test statistic columns represent independent t statistics for continuous variables and χ² statistics for categorical variables. For the postmatched cohort, all statistics and P values were derived using paired tests appropriate for matched samples, specifically the paired t test for continuous variables and the McNemar test for categorical variables. An absolute standardized mean difference less than 0.10 indicates negligible imbalance between groups. FTS: Fast track surgery; BMI: Body mass index; NRS 2002: Nutritional Risk Screening 2002; HADS: Hospital Anxiety and Depression Scale; ASA: American Society of Anesthesiologists; CCI: Charlson Comorbidity Index; AJCC: American Joint Committee on Cancer; SMD: Standardized mean difference.

Table 2 Comparison of primary outcomes in the matched cohort (n = 250), n (%)/mean ± SD

Outcome	Watson-FTS (n = 125)	Routine care (n = 125)	Unadjusted difference	Adjusted effect (95%CI)1	P value
First flatus time (hours)	52.4 ± 10.8	68.1 ± 13.5	-15.7	β: -15.2 (-18.4 to -12.0)	< 0.001
Overall complications	19 (15.2)	38 (30.4)	-15.20	aOR: 0.39 (0.21-0.72)	0.003
Clavien-Dindo severity
Grade I-II	15 (12.0)	30 (24.0)	-12.00	aOR: 0.43 (0.22-0.85)	0.015
Grade ≥ III	4 (3.2)	8 (6.4)	-3.20	aOR: 0.48 (0.14-1.66)	0.246
Stoma-related events	6 (4.8)	16 (12.8)	-8.00	aOR: 0.34 (0.13-0.89)	0.024

¹Propensity score matching was performed using all baseline variables listed in Table 1 (including age, sex, body mass index, education, American Society of Anesthesiologists, Nutritional Risk Screening 2002, Charlson Comorbidity Index, smoking, neoadjuvant therapy, Hospital Anxiety and Depression Scale anxiety/depression, surgical approach, TNM stage, tumor location, permanent stoma status, and baseline nutritional markers). The regression models for outcomes further adjusted for these covariates to account for residual confounding.

Statistical significance was set at P < 0.05. β: Adjusted regression coefficient from multivariable linear regression. aOR: Adjusted odds ratio from multivariable logistic regression. FTS: Fast track surgery; aOR: Adjusted odds ratio; CI: Confidence interval.

Table 3 Comparison of secondary outcomes in the matched cohort (n = 250), n (%)/mean ± SD

Outcome dimension	Watson-FTS (n = 125)	Routine care (n = 125)	Unadjusted difference	Adjusted effect (95%CI)1	P value
Physiological and functional
First ambulation time (hours)	26.4 ± 6.2	38.9 ± 8.5	-12.5	β: -12.1 (-13.8 to -10.4)	< 0.001
Postoperative stay (days)2, median (IQR)	7.0 (6.0, 8.0)	9.0 (7.5, 11.0)	-2.0	β: -1.8 (-2.3 to -1.3)	< 0.001
Discharge KPS score	84.5 ± 5.2	78.2 ± 6.1	+6.3	β: +5.9 (4.6-7.2)	< 0.001
Humanistic process quality
Teach-back completion	115 (92.0)	81 (64.8)	+27.2	aOR: 5.82 (2.76-12.28)	< 0.001
Individualized plan rate	111 (88.8)	65 (52.0)	+36.8	aOR: 7.14 (3.58-14.24)	< 0.001
Self-efficacy milestone3	102 (81.6)	71 (56.8)	+24.8	aOR: 3.25 (1.84-5.74)	< 0.001

¹All models were adjusted for all baseline covariates listed in Table 1 to account for residual confounding.

²P value for length of stay derived from Wilcoxon signed-rank test; β derived from generalized linear model with log-link to account for skewness.

³Defined as the documented independent performance of a complete stoma appliance change by the patient prior to discharge.

β: Adjusted regression coefficient. KPS: Karnofsky Performance Status; FTS: Fast track surgery; aOR: Adjusted odds ratio; CI: Confidence interval.

Table 4 Generalized estimating equation model results for nutritional trajectories and protocol adherence (n = 250)

Model parameter	Prealbumin (mg/L) β (95%CI)	P value	FTS adherence (yes) OR (95%CI)	P value
Intercept	222.8 (215.4-230.2)	< 0.001	0.82 (0.54-1.25)	0.358
Group (Watson-FTS vs control)	5.48 (2.12-8.84)	0.001	2.15 (1.42-3.26)	< 0.001
Time (categorical reference: T1)
T2 (day 3 for nutrition/adherence)	17.6 (12.4-22.8)	< 0.001	2.56 (1.88-3.48)	< 0.001
T3 (day 7 for nutritional/day 5 for adherence)	33.7 (28.5-38.9)	< 0.001	4.12 (2.95-5.76)	< 0.001
Group × time interaction	6.42 (4.15-8.69)	< 0.001	1.85 (1.18-2.90)	0.008

Data from propensity score matching-matched cohort (n = 125 pairs). Generalized estimating equation models utilized an exchangeable correlation structure. All models were adjusted for the baseline values of the respective outcome and the comprehensive set of covariates listed in Table 1 (including age, body mass index, American Society of Anesthesiologists class, TNM stage, and surgical approach) to account for residual confounding. Interaction odds ratio > 1 or β > 0 indicates a more favorable trajectory in the Watson-fast track surgery group. FTS: Fast track surgery; OR: Odds ratio; CI: Confidence interval.

Table 5 Multivariable regression analysis for predictors of postoperative recovery quality (n = 312)

Predictors	Model 1 early recovery success (flatus < 48 hours)			Model 2 functional status at discharge (KPS)
	Adjusted OR (95%CI)	χ2	P value	Adjusted β (95%CI)	t value	P value
Watson-FTS intervention (ref: Routine)	3.62 (2.18-6.01)	24.15	< 0.001	6.24 (4.85-7.63)	8.82	< 0.001
Age (per year)	0.97 (0.94-0.99)	5.02	0.025	-0.08 (-0.16 to 0.01)	-2.14	0.033
BMI (kg/m2)	1.04 (0.96-1.13)	0.88	0.348	0.12 (-0.15 to 0.39)	0.86	0.391
Laparoscopic approach (ref: Open)	2.45 (1.34-4.48)	8.24	0.004	2.85 (1.12-4.58)	3.24	0.001
ASA class III (ref: I-II)	0.78 (0.45-1.35)	0.76	0.383	-1.45 (-3.12 to 0.22)	-1.71	0.088
TNM stage III (ref: I-II)	0.82 (0.51-1.32)	0.64	0.424	-3.12 (-5.45 to 0.79)	-2.63	0.009
Permanent stoma (ref: Temporary)	0.88 (0.54-1.43)	0.28	0.597	-2.05 (-3.82 to 0.28)	-2.28	0.023

Model 1 utilized multivariable logistic regression (Hosmer-Lemeshow, P = 0.742). Model 2 utilized multivariable linear regression (adjusted R squared = 0.485). Reference groups are indicated in parentheses.

KPS: Karnofsky Performance Status; BMI: Body mass index; ASA: American Society of Anesthesiologists; OR: Odds ratio; β: Unstandardized regression coefficient; CI: Confidence interval.

Table 6 Subgroup analysis of the effect of Watson-fast track surgery on time to first flatus

Subgroup	n (Int/Con)	Adjusted beta (95%CI)	P value	P for interaction
Overall matched cohort	125/125	-15.2 (-18.4 to -12.0)	< 0.001	Reference
Age (years)				0.784
< 65	74/76	-15.4 (-18.9 to -11.9)	< 0.001
≥ 65	51/49	-14.8 (-19.2 to -10.4)	< 0.001
Surgical approach				0.412
Laparoscopic	95/94	-15.6 (-19.1 to -12.1)	< 0.001
Open	30/31	-13.9 (-18.2 to -9.6)	< 0.001
Stoma type				0.521
Permanent	50/49	-16.2 (-19.8 to -12.6)	< 0.001
Temporary	75/76	-14.5 (-18.1 to -10.9)	< 0.001

Data are derived from the propensity score-matched cohort. Int/Con: Intervention group (Watson-fast track surgery)/control group (routine care). The adjusted beta represents the mean difference in time to first postoperative flatus (in hours), estimated via multivariable linear regression adjusting for baseline covariates. A negative Beta indicates a shorter recovery time for the Watson-fast track surgery group. The P for interaction assesses whether the magnitude of the intervention effect differs significantly across the strata of each subgroup. CI: Confidence interval.

Table 7 Interrupted time-series analysis (segmented linear regression) evaluating the impact of the Watson-fast track surgery intervention on time to first flatus

Parameter	Estimate (hours)	SE	95%CI	t	P value
Baseline intercept (estimated time at Q1 2019)	68.52	1.84	64.91-72.13	37.24	< 0.001
Preintervention trend (change per quarter before Jan 2022)	-0.15	0.14	-0.42 to 0.12	-1.07	0.312
Level change (intervention effect) (immediate drop at Q1 2022)	-13.84	1.88	-17.52 to -10.16	-7.36	< 0.001
Postintervention trend modification (change in slope after intervention)	-0.05	0.1	-0.24 to 0.14	-0.5	0.62

The segmented linear regression model utilized aggregated quarterly data from January 2019 to December 2024. A negative estimate for the “Level Change” indicates a robust, immediate acceleration in gastrointestinal recovery (shorter time to flatus) exactly coinciding with the implementation of the Watson-fast track surgery pathway, independent of the nonsignificant underlying secular trends. CI: Confidence interval.