Association of Dementia With Total Ankle Arthroplasty and Ankle Arthrodesis Outcomes: A Short Report

Abstract

Background:

Foot and ankle surgeons evaluate older patients with ankle arthritis and dementia. In this report, we analyzed postoperative complications following ankle arthrodesis (AA) and total ankle arthroplasty (TAA), hypothesizing greater risk in patients with dementia.

Methods:

We conducted a retrospective cohort study using Medicare claims from 2016 to 2021, including beneficiaries aged ≥65 years who underwent TAA or AA. The primary independent variable was dementia, defined using the Charlson Comorbidity Index (CCI) dementia component based on International Classification of Diseases, Tenth Revision (ICD-10) diagnosis codes within 90 days before the index procedure. Outcomes included postoperative complications, all-cause readmissions, and mortality within 90 days and 12 months. Multivariable logistic regression models estimated adjusted differences in complication rates by dementia status.

Results:

Dementia was present in 0.56% of TAA patients (73 of 13 134) and 3.6% of AA patients (308 of 8519). With the numbers available, no significant differences could be detected with dementia in TAA using adjusted multivariable analyses, although 12-month readmissions were higher (P = .046). In contrast, among AA patients, dementia was associated with increased 90-day readmissions (P = .041) and higher 12-month mortality (P = .001), whereas other outcomes were not statistically significant.

Conclusion:

We found differences in postoperative outcomes, particularly readmissions and mortality, in patients with dementia following AA. Conversely, we found limited differences in patients with and without dementia undergoing TAA, which may reflect limited power to detect modest differences given the smaller number of dementia patients in the TAA cohort. Further research should explore how the severity of cognitive impairment influences decision making in ankle arthritis treatment.

Level of Evidence:

Level III, retrospective cohort study.

Introduction

Surgical treatment options for ankle arthritis include ankle arthrodesis (AA) and total ankle arthroplasty (TAA). ¹ Both operations are commonly performed in older patients. ² Similarly, in an aging population, the number of adults in the United States who will develop dementia is projected to increase from 514 000 in 2020 to roughly 1 million in 2060. ³ Accordingly, foot and ankle surgeons will increasingly evaluate older patients with both symptomatic ankle arthritis and dementia. There is currently little evidence evaluating the impact of dementia on outcomes following AA or TAA. In this short report, we analyzed postoperative complication rates following AA and TAA in a Medicare cohort, hypothesizing greater risk in patients with dementia than in patients without dementia.

Methods

We conducted a retrospective cohort study using Medicare fee-for-service (FFS) and Medicare Advantage (health maintenance organization [HMO]) claims from 2016 to 2021, including beneficiaries aged ≥65 years who underwent TAA or AA. Surgeries were identified using relevant Current Procedural Terminology (CPT) codes (TAA: 27700, 27702; AA: 27870, 29899), and the index procedure was defined as the first qualifying TAA or AA during the study period. Patients were required to have continuous enrollment during the 90-day preoperative look-back period to ensure complete capture of comorbidities. Beneficiaries eligible for Medicare through end-stage renal disease (ESRD), or Social Security Disability Insurance (SSDI), were excluded.

The primary independent variable was dementia status, defined using the Charlson Comorbidity Index (CCI) dementia component based on International Classification of Diseases, Tenth Revision (ICD-10) diagnosis codes within 90 days prior to the index procedure. Outcomes included postoperative complications, all-cause readmissions, and mortality within 90 days and 12 months following discharge. Continuous postoperative enrollment through the corresponding follow-up period was required for ascertainment of 90-day and 12-month outcomes. Complications were identified using inpatient and part B claims and categorized using previously published ICD-10–based algorithms. ⁴ Mortality was assessed using the Master Beneficiary Summary File and analyzed separately. Covariates included age (5-year increments), sex, race, obesity, HMO status, and all individual CCI components (excluding dementia) as separate covariates in the primary model. A clinically targeted model including diabetes, peripheral vascular disease, and chronic kidney disease as well as a grouped CCI model were also evaluated as sensitivity analyses.

We compared unadjusted complication rates between patients with and without dementia separately for TAA or AA. Chi-square tests were used for categorical variables and Welch t test for continuous outcomes. Multivariable logistic regression models estimated adjusted differences in complication rates by dementia status, reported as percentage-point (ppt) differences with 95% CIs. Year fixed effects were included, and SEs were clustered at the hospital level. To further address potential selection bias, inverse probability-weighted regression adjustment (IPWRA) was performed as a sensitivity analysis. Additional sensitivity analyses were conducted using alternative comorbidity specifications, including selected clinically relevant comorbidities and grouped CCI models. Post hoc minimum detectable effects were calculated assuming 80% power and a 2-sided α of 0.05. All analyses were conducted in STATA-MP version 19.5 within the CMS Virtual Research Data Center (VRDC).

Results

Dementia was present in 0.56% of TAA patients (73 of 13 134) and 3.6% of AA patients (308 of 8519). In both cohorts, patients with dementia were older and had a higher burden of comorbid conditions than those without dementia (Table 1).

Table 1.

Descriptive Statistics for TAA and AA by Dementia Status. ^a

Variables	TAA					AA
	Non-dementia (N = 13 061)		Dementia (N = 73)		P Value	Non-dementia (n = 8211)		Dementia (n = 308)		P Value
Age, mean (SD)	72.8	(5.2)	76.5	(6.7)	<.001	73.4	(5.9)	80.1	(8.4)	<0.001
Age group (5-y increments)
65-69	4188	32.1%	14	19.2%	<.001	2515	30.6%	38	12.3%	<.001
70-74	4412	33.8%	14	19.2%		2639	32.1%	55	17.9%
75-79	2856	21.9%	22	30.1%		1823	22.2%	53	17.2%
80-84	1271	9.7%	Suppressed			804	9.8%	59	19.2%
85+	325	2.5%	Suppressed			430	5.2%	103	33.4%
Female					.093					<.001
No	7311	56.0%	48	65.7%		4188	51.0%	89	28.9%
Yes	5750	44.0%	25	34.3%		4023	49.0%	219	71.1%
Race (3 groups)					.550					.004
White	12272	94.0%	69	94.5%		7487	91.2%	273	88.6%
Black	200	1.5%	Suppressed			331	4.0%	24	7.8%
Other	589	4.5%	Suppressed			393	4.8%	11	3.6%
Charlson Index: 90-d lookback					<.001					<.001
0	6247	53.1%	0	0%		2670	32.5%	0	0%
1	2541	21.6%	23	31.5%		1642	20.0%	36	11.7%
2	2988	25.4%	50	68.5%		3899	47.5%	272	88.3%
Peripheral vascular					<.001					<.001
No	11823	90.5%	50	68.5%		6759	82.3%	199	64.6%
Yes	1238	9.5%	23	31.5%		1452	17.7%	109	35.4%
Diabetes					.825					.002
No	11037	84.5%	61	84.6%		5425	66.1%	177	57.5%
Yes	2024	15.5%	12	16.4%		2786	33.9%	131	42.5%
Chronic kidney disease					.001					<.001
No	11963	91.6%	59	80.8%		6643	80.9%	206	66.9%
Yes	1098	8.4%	14	19.2%		1568	19.1%	102	33.1%
Obesity					.773					.020
No	12238	93.7%	Suppressed			7022	85.5%	278	90.3%
Yes	823	6.3%	Suppressed			1189	14.5%	30	9.7%
HMO					.851					.146
No	11891	91.0%	Suppressed			7357	89.6%	268	87.0%
Yes	1170	9.0%	Suppressed			854	10.4%	40	13.0%

Abbreviations: AA, ankle arthroplasty; HMO, health maintenance organization; TAA, total ankle arthroplasty.

^aSome data were suppressed because of small cell size (<10 patients) in accordance with CMS privacy policy.

In adjusted multivariable analyses (Table 2) with the numbers available, dementia was not significantly associated with 90-day or 12-month complications following TAA. However, TAA patients with dementia had higher 12-month readmissions (+11.7 ppt; 95% CI 0.2-23.2; P = .046). Among AA patients, dementia was associated with higher 90-day readmissions (+5.0 ppt; P = .041) and higher 12-month mortality (+4.9 ppt; P = .001). Other outcomes were not statistically significant. TAA DVT estimates were not reported because sparse events prevented reliable estimation.

Table 2.

Adjusted Differences in Complication Rates Following TAA and AA Among Medicare Beneficiaries by Dementia Status. ^a

Complication outcomes		(a) TAA		(b) AA
		Logistic Regression
		Difference, ppt (95% CI)	P Value	Difference, ppt (95% CI)	P Value
90 d
Any complication	Non-dementia	Ref	–	Ref	–
	Dementia	8.4 (−2.2; 19.1)	.121	4.4 (−2.0; 10.9)	.180
Infection	Non-dementia	Ref	–	Ref	–
	Dementia	1.4 (−3.3; 6.1)	.549	3.3 (−0.2; 6.7)	.061
PE	Non-dementia	Ref	–	Ref	–
	Dementia	0.2 (−1.6; 2.0)	.853	−0.1 (−1.5; 1.5)	.109
DVT	Non-dementia	–	–	Ref	–
	Dementia	–	–	−0.3 (−1.8; 1.3)	.733
Device	Non-dementia	Ref	–	Ref	–
	Dementia	1.4 (−6.4; 9.2)	.726	0.6 (−0.3: 1.5)	.175
Mortality	Non-dementia	Ref	–	Ref	–
	Dementia	0.6 (−1.0; 2.1)	.477	0.5 (−0.3; 1.2)	.215
Readmission	Non-dementia	Ref	–	Ref	–
	Dementia	7.1 (−0.1; 14.1)	.051	5.0 (0.2; 9.8)	.041
12 mo
Any complication	Non-dementia	Ref	–	Ref	–
	Dementia	2.5 (−9.3; 14.2)	.679	0.2 (−7.2; 7.6)	.955
Infection	Non-dementia	Ref	–	Ref	–
	Dementia	2.7 (−3.5; 8.9)	.392	2.8 (−1.6; 7.1)	.214
PE	Non-dementia	Ref	–	Ref	–
	Dementia	−0.1 (−2.2; 1.9)	.904	−0.3 (−2.1; 1.5)	.768
DVT	Non-dementia	–	–	Ref	–
	Dementia	–	–	−0.6 (−3.2; 1.9)	.645
Device	Non-dementia	Ref	–	Ref	–
	Dementia	2.6 (−7.0: 12.2)	.593	−3.5 (−10.2; 3.2)	.306
Mortality	Non-dementia	Ref	–	Ref	–
	Dementia	0.1 (−1.5; 1.6)	.929	4.9 (2.0; 7.8)	.001
Readmission	Non-dementia	Ref	–	Ref	–
	Dementia	11.7 (0.2; 23.2)	.046	7.2 (−0.2; 14.5)	.057

Abbreviations: AA, ankle arthroplasty; DVT, deep vein thrombosis; PE, pulmonary embolism; ppt, percentage point; TAA, total ankle arthroplasty.

^aModels adjusted for age, sex, race, health maintenance organization (HMO) status, and all individual Charlson Comorbidity Index components (excluding dementia). Dementia was defined using the Charlson dementia component. TAA DVT estimates were not reported because sparse events prevented reliable estimation.

In IPWRA models (Table 3), dementia was not associated with increased complications in TAA patients. In contrast, among AA patients, dementia was associated with higher 12-month mortality (+7.1 ppt; 95% CI 1.9-12.4; P = .007) and higher rates of DVT (+3.1 ppt; 95% CI 1.1-5.0; P = .002). Although the direction-of-effect was generally similar to the primary analyses, specific outcomes reaching statistical significance differed across model specifications.

Table 3.

Matched Comparison of Complication Rates After TAA and AA in Medicare Beneficiaries by Dementia Status. ^a

Complication outcomes		(a) TAA		(b) AA
		Matched Cohort
		Difference, ppt (95% CI)	P Value	Difference, ppt (95% CI)	P Value
90 d  Any complication
	Non-dementia	Ref	–	Ref	–
	Dementia	8.7 (−8.8; 26.1)	.331	10.3 (−2.1; 22.6)	.103
Infection	Non-dementia	Ref	–	Ref	–
	Dementia	−0.3 (−3.8; 3.3)	.885	6.0 (−3.3; 15.2)	.205
PE	Non-dementia	Ref	–	Ref	–
	Dementia	−0.3 (−1.1; 0.6)	.562	−0.6 (−1.6; 0.4)	.261
DVT	Non-dementia	–	–	Ref	–
	Dementia	–	–	0.4 (−2.9: 3.6)	.832
Device	Non-dementia	Ref	–	Ref	–
	Dementia	3.1 (−8.8: 15.0)	.609	11.1 (−0.1: 23.0)	.072
Mortality	Non-dementia	Ref	–	Ref	–
	Dementia	0.3 (−0.8; 1.5)	.541	0.7 (−0.8; 2.1)	.373
Readmission	Non-dementia	Ref	–	Ref	–
	Dementia	3.6 (−4.3; 11.6)	.373	7.5 (−2.8; 17.8)	.155
12 mo
Any complication	Non-dementia	Ref	–	Ref	–
	Dementia	7.3 (−13.7;28.3)	.498	1.2 (−12.8; 15.4)	.858
Infection	Non-dementia	Ref	–	Ref	–
	Dementia	−1.2 (−4.9; 2.5)	.518	8.6 (−3.1; 20.2)	.149
PE	Non-dementia	Ref	–	Ref	–
	Dementia	−0.7 (−1.6; 0.2)	.120	0.6 (−2.0; 3.1)	.667
DVT	Non-dementia	–	–	Ref	–
	Dementia	–	–	3.1 (1.1; 5.0)	.002
Device	Non-dementia	Ref	–	Ref	–
	Dementia	8.7 (−7.3: 24.8)	.287	6.1 (−7.8: 19.9)	.391
Mortality	Non-dementia	Ref	–	Ref	–
	Dementia	−0.4 (−1.5; 0.7)	.491	7.1 (1.9; 12.4)	.007
Readmission	Non-dementia	Ref	–	Ref	–
	Dementia	4.1 (−10.6; 18.9)	.583	12.5 (−1.5; 26.5)	.080

Abbreviations: AA, ankle arthroplasty; DVT, deep vein thrombosis; PE, pulmonary embolism; ppt, percentage point; TAA, total ankle arthroplasty.

^aNote: Matched analyses used the same covariate specifications as the corresponding regression models. TAA DVT estimates were not reported because sparse events prevented reliable estimation.

Results were also similar in direction across alternative model specifications using selected comorbidities and grouped CCI (Supplementary Tables 1 and 2), although statistical significance differed across models. Because baseline event rates vary by outcome, the study’s detectable effect size differs across endpoints. For the 90-day any-complication outcome, the minimum detectable absolute difference (80% power, 2-sided α = 0.05) was approximately 8.2 ppt for TAA and 6.9 ppt for AA, suggesting limited power to detect modest differences across several outcomes, particularly within the smaller TAA dementia subgroup (see Supplementary Table 3).

Discussion

Here, we studied the role of dementia on complications following elective TAA and AA. The TAA and AA cohorts were analyzed as separate clinical populations, and no direct comparison between procedures was intended. Dementia was associated with significantly higher rates of readmission and mortality following AA. In contrast, among TAA patients, no consistent association was observed across outcomes with the numbers available, although higher 12-month readmissions were noted. However, the small number of dementia patients undergoing TAA limits the ability to detect modest differences. Across multiple modeling approaches, the direction of effect was generally consistent, with no clear association observed in TAA and a signal toward increased adverse outcomes in AA. However, the specific outcomes reaching statistical significance varied across models, likely reflecting differences in model specification and limited power in certain subgroups.

Prior studies have analyzed the risk of dementia on elective orthopaedic surgery. Serna et al ⁵ showed greater risk of periprosthetic fracture, instability, and need for revision in patients with dementia undergoing elective shoulder replacements. Dementia was independently associated with an increased risk of perioperative complications following elective TKA ⁶ and THA, ⁷ but other reports highlighted marginal differences. ⁸ Dementia was associated with increased postoperative complications within 30 days following elective cervical and lumbar spine surgery. ⁹ Lee et al ¹⁰ showed that dementia was a possible risk factor for early TAA failure, but beyond that, no other studies have examined dementia in either TAA or AA.

There are clear limitations of this study. The claims-based definition for dementia likely captures a heterogeneous spectrum of cognitive impairment, and we were unable to assess dementia severity for an individual patient. In addition, dementia was defined using administrative claims and does not allow differentiation by severity or subtype. Although subgroup analyses using specific diagnosis codes may be possible, sample sizes would be small and unlikely to yield reliable estimates. Dementia rates (TAA 0.56%, AA 3.6%) were low, particularly in the TAA cohort, so we may have been underpowered to detect modest differences, particularly in the TAA cohort. But the dementia rates here are similar to elective shoulder, hip, and knee arthroplasties.^5-7 Because this study is observational, we cannot make definitive causal inferences about the relationship between dementia and complications following both AA and TAA surgeries. In addition, as with all claims-based analyses, residual confounding from unmeasured clinical factors may influence the observed associations. Administrative data also lack detailed clinical information, which limits risk adjustment and outcome interpretation. Further, our analyses are exploratory and hypothesis-generating but not confirmatory. Although we found associations of higher risk in patients with dementia, these findings cannot be interpreted as causal. Lastly, this study was not designed to compare the risk profiles between AA and TAA. Although our results suggest higher complications in dementia patients for AA, these findings do not necessarily indicate that TAA is safer than AA. Patients with dementia may be preferentially selected for AA rather than TAA, and this potential selection bias may partially explain the higher complication rates observed in the AA cohort.

Conclusion

In this study, we found notable differences in postoperative readmissions and mortality in patients with dementia following AA. In the TAA cohort, we found limited differences in patients with and without dementia, which may in part reflect limited power to detect modest differences given the smaller number of dementia patients in the TAA cohort. Further research should explore how the severity of cognitive impairment influences ankle arthritis treatment.

Supplementary Table 1.

(a) Adjusted Differences in Complication Rates Following TAA and AA Among Medicare Beneficiaries by Dementia Status (With Diabetes, Peripheral Vascular Disease, Chronic Kidney Disease, and Obesity in the Model Instead of CCI). ^a

Complication Outcomes		(a) TAA		(b) AA
		Logistic Regression
		Difference, ppt (95% CI)	P Value	Difference, ppt (95% CI)	P Value
90 d  Any complication
	Non-dementia	Ref	–	Ref	–
	Dementia	9.8 (−0.1; 20.1)	.076	6.2 (−0.2; 12.6)	.059
Infection	Non-dementia	Ref	–	Ref	–
	Dementia	1.4 (−3.3; 6.2)	.554	3.6 (0.1; 7.2)	.044
PE	Non-dementia	Ref	–	Ref	–
	Dementia	0.2 (−1.8; 2.1)	.863	−0.6 (−1.6; 0.4)	.251
DVT	Non-dementia	–	–	Ref	–
	Dementia	–	–	−0.1 (−1.7; 1.7)	.987
Device	Non-dementia	Ref	–	Ref	–
	Dementia	1.5 (−6.4; 9.4)	.703	1.8 (−3.6: 7.3)	.508
Mortality	Non-dementia	Ref	–	Ref	–
	Dementia	0.5 (−0.9; 1.9)	.489	0.7 (−0.3; 1.7)	.167
Readmission	Non-dementia	Ref	–	Ref	–
	Dementia	6.9 (−0.1; 13.8)	.054	6.2 (1.3; 11.1)	.014
12 mo
Any complication	Non-dementia	Ref	–	Ref	–
	Dementia	3.9 (−8.0; 15.9)	.517	2.2 (−5.0; 9.3)	.558
Infection	Non-dementia	Ref	–	Ref	–
	Dementia	2.7 (−3.5; 9.0)	.393	3.3 (−1.2; 7.8)	.149
PE	Non-dementia	Ref	–	Ref	–
	Dementia	−0.1 (−2.4; 2.3)	.969	−0.1 (−2.1; 2.0)	.972
DVT	Non-dementia	–	–	Ref	-
	Dementia	–	–	−0.4 (−3.1; 2.3)	.784
Device	Non-dementia	Ref	–	Ref	–
	Dementia	2.8 (−6.9: 12.5)	.567	−3.5 (−1.0: 3.2)	.302
Mortality	Non-dementia	Ref	–	Ref	–
	Dementia	0.1 (−1.7; 1.9)	.926	5.2 (2.2; 8.3)	.001
Readmission	Non-dementia	Ref	–	Ref	–
	Dementia	12.6 (1.0; 24.3)	.033	8.9 (1.6; 16.2)	.017

Abbreviations: AA, ankle arthroplasty; CCI, Charlson Comorbidity Index; DVT, deep vein thrombosis; PE, pulmonary embolism; ppt, percentage point; TAA, total ankle arthroplasty.

^aModels adjusted for age, sex, race, diabetes, peripheral vascular disease, chronic kidney disease, obesity, surgical setting, provider type, and health maintenance organization (HMO) status. TAA DVT estimates were not reported because sparse events prevented reliable estimation.

(b) Adjusted differences in Complication Rates Following TAA and AA Among Medicare Beneficiaries by Dementia Status (With Grouped CCI [0, 1, 2+], Without Dementia Status, in the Model). ^a

Complication Outcomes		(a) TAA		(b) AA
		Logistic Regression
		Difference, ppt (95% CI)	P Value	Difference, ppt (95% CI)	P Value
90 d  Any complication
	Non-dementia	Ref	–	Ref	–
	Dementia	9.1 (−1.6; 19.9)	.096	7.0 (0.6; 13.4)	.032
Infection	Non-dementia	Ref	–	Ref	–
	Dementia	1.6 (−3.2; 6.5)	.510	4.0 (0.4; 7.6)	.029
PE	Non-dementia	Ref	–	Ref	–
	Dementia	0.2 (−1.8; 2.1)	.875	−0.6 (−1.6; 0.4)	.207
DVT	Non-dementia	–	–	Ref	–
	Dementia	–	–	−0.1 (−1.6; 1.6)	.985
Device	Non-dementia	Ref	–	Ref	–
	Dementia	1.6 (−6.3; 9.5)	.689	2.5 (−3.1; 8.0)	.380
Mortality	Non-dementia	Ref	–	Ref	–
	Dementia	0.5 (−0.9; 1.8)	.493	0.8 (−0.3; 1.8)	.143
Readmission	Non-dementia	Ref	–	Ref	–
	Dementia	6.9 (−0.1; 13.8)	.051	6.6 (1.7; 11.4)	.009
12 mo
Any complication	Non-dementia	Ref	–	Ref	–
	Dementia	4.6 (−7.4; 16.6)	.449	2.7 (−4.4; 9.8)	.457
Infection	Non-dementia	Ref	–	Ref	–
	Dementia	3.0 (−3.4; 9.4)	.357	3.7 (−0.9; 8.2)	.115
PE	Non-dementia	Ref	–	Ref	–
	Dementia	−0.1 (−2.5; 1.2)	.924	−0.1 (−2.1; 1.8)	.893
DVT	Non-dementia	–	–	Ref	–
	Dementia	–	–	−0.5 (−3.0; 2.1)	.724
Device	Non-dementia	Ref	–	Ref	–
	Dementia	3.1 (−6.7; 12.9)	.531	−3.0 (−9.8; 3.7)	.374
Mortality	Non-dementia	Ref	–	Ref	–
	Dementia	0.1 (−1.5; 1.6)	.954	5.5 (2.4; 8.9)	<.001
Readmission	Non-dementia	Ref	–	Ref	–
	Dementia	12.7 (1.0; 24.3)	.033	8.7 (1.5; 15.8)	.018

Abbreviations: AA, ankle arthroplasty; CCI, Charlson Comorbidity Index; DVT, deep vein thrombosis; PE, pulmonary embolism; ppt, percentage point; TAA, total ankle arthroplasty.

^aModels adjusted for age, sex, race, grouped Charlson Comorbidity Index (0, 1, ≥2), excluding dementia, obesity, surgical setting, provider type, and health maintenance organization (HMO) status. TAA DVT estimates were not reported because sparse events prevented reliable estimation.

Supplementary Table 2.

(a) Matched Comparison of Complication Rates After TAA and AA in Medicare Beneficiaries by Dementia Status (With Diabetes, Peripheral Vascular Disease, Chronic Kidney Disease, and Obesity, Instead of CCI). ^a

Complication Outcomes		(a) TAA		(b) AA
		Matched Cohort
		Difference, ppt (95% CI)	P Value	Difference, ppt (95% CI)	P Value
90 d  Any complication
	Non-dementia	Ref	–	Ref	–
	Dementia	9.2 (−6.5; 25.0)	.249	10.8 (1.1; 20.6)	.029
Infection	Non-dementia	Ref	–	Ref	–
	Dementia	0.9 (−6.9; 8.8)	.820	3.4 (−2.7; 9.6)	.275
PE	Non-dementia	Ref	–	Ref	–
	Dementia	0.1 (−1.2; 1.2)	.977	−0.5 (−1.7; 0.6)	.378
DVT	Non-dementia	–	–	Ref	–
	Dementia	–	–	1.6 (−1.2; 4.4)	.250
Device	Non-dementia	Ref	–	Ref	–
	Dementia	3.5 (−8.1; 15.1)	.557	5.3 (−3.4; 14.1)	.231
Mortality	Non-dementia	Ref	–	Ref	–
	Dementia	0.3 (−0.6; 1.2)	.523	2.6 (−0.7; 6.0)	.122
Readmission	Non-dementia	Ref	–	Ref	–
	Dementia	4.8 (−3.6; 13.2)	.263	7.9 (0.2; 15.9)	.048
12 mo
Any complication	Non-dementia	Ref	–	Ref	–
	Dementia	5.7 (−12.7; 24.2)	.541	10.6 (0.9; 20.3)	.032
Infection	Non-dementia	Ref	–	Ref	–
	Dementia	1.3 (−6.4; 9.2)	.738	4.2 (−3.1; 11.5)	.255
PE	Non-dementia	Ref	–	Ref	–
	Dementia	−0.4 (−1.7; 1.0)	.582	1.3 (−3.2; 5.9)	.566
DVT	Non-dementia	–	–	Ref	–
	Dementia	–	–	0.2 (−4.2; 4.5)	.946
Device	Non-dementia	Ref	–	Ref	–
	Dementia	6.7 (−7.6; 21.0)	.361	1.2 (−7.1; 9.4)	.781
Mortality	Non-dementia	Ref	–	Ref	–
	Dementia	−0.4 (−1.4; 0.5)	.369	9.9 (4.2; 15.7)	.001
Readmission	Non-dementia	Ref	–	Ref	–
	Dementia	7.3 (−5.4; 20.7)	.260	11.3 (1.1; 21.4)	.029

Abbreviations: AA, ankle arthroplasty; CCI, Charlson Comorbidity Index; DVT, deep vein thrombosis; PE, pulmonary embolism; ppt, percentage point; TAA, total ankle arthroplasty.

^aMatched analyses used the same covariate specifications as the corresponding regression models. TAA DVT estimates were not reported because sparse events prevented reliable estimation.

(b) Matched Comparison of Complication Rates After TAA and AA in Medicare Beneficiaries by Dementia Status (With Grouped CCI [0, 1, 2+], Without Dementia Status, in the Model). ^a

Complication Outcomes		(a) TAA		(b) AA
		Matched Cohort
		Difference, ppt (95% CI)	P Value	Difference, ppt (95% CI)	P Value
90 d  Any complication
	Non-dementia	Ref	–	Ref	–
	Dementia	8.4 (−4.2; 21.1)	.190	13.1 (4.4; 21.7)	.003
Infection	Non-dementia	Ref	–	Ref	–
	Dementia	−0.1 (−6.3; 6.0)	.970	5.3 (−0.2; 10.8)	.057
PE	Non-dementia	Ref	–	Ref	–
	Dementia	0.1 (−2.7; 2.9)	.962	−0.5 (−2.5; 1.4)	.587
DVT	Non-dementia	–	–	Ref	–
	Dementia	–	–	2.9 (−0.9: 6.8)	.135
Device	Non-dementia	Ref	–	Ref	–
	Dementia	2.1 (−6.9; 11.2)	.644	5.7 (−1.7; 13.2)	.133
Mortality	Non-dementia	Ref	–	Ref	–
	Dementia	0.1 (−0.1; 1.1)	.781	3.4 (0.2; 6.6)	.040
Readmission	Non-dementia	Ref	–	Ref	–
	Dementia	10.0 (0.1; 19.1)	.032	14.0 (6.1; 21.9)	.001
12 mo
Any complication	Non-dementia	Ref	–	Ref	–
	Dementia	4.8 (−11.5; 21.1)	.563	11.2 (2.1; 20.3)	.016
Infection	Non-dementia	Ref	–	Ref	–
	Dementia	1.4 (−4.7; 7.5)	.649	4.8 (−1.3; 10.9)	.120
PE	Non-dementia	Ref	–	Ref	–
	Dementia	−0.3 (−3.5; 2.9)	.842	1.2 (−2.5; 4.9)	.522
DVT	Non-dementia	–	–	Ref	–
	Dementia	–	–	2.6 (−2.9; 8.1)	.357
Device	Non-dementia	Ref	–	Ref	–
	Dementia	5.0 (−7.1; 17.1)	.417	1.0 (−7.9; 9.9)	.828
Mortality	Non-dementia	Ref	–	Ref	–
	Dementia	−0.6 (−1.4; 0.1)	.105	11.2 (5.2; 17.3)	<.001
Readmission	Non-dementia	Ref	–	Ref	–
	Dementia	13.2 (1.7; 24.8)	.025	15.0 (5.2; 24.8)	.003

Abbreviations: AA, ankle arthroplasty; CCI, Charlson Comorbidity Index; DVT, deep vein thrombosis; PE, pulmonary embolism; ppt, percentage point; TAA, total ankle arthroplasty.

^aMatched analyses used the same covariate specifications as the corresponding regression models. TAA DVT estimates were not reported because sparse events prevented reliable estimation.

Supplementary Table 3.

Post Hoc Minimum Detectable Effects (80% Power, 2-Sided α = 0.05) for Complication Outcomes Based on Observed Group Sizes. ^a

Outcome	MDE (ppt)
	TAA		AA
	90-d	12-mo	90-d	12-mo
Any complication	8.21	6.30	6.89	6.72
Infection	4.67	5.07	4.97	6.59
PE	2.81	2.99	3.06	3.98
DVT	3.44	4.05	4.53	5.98
Device problems	6.89	6.01	7.12	6.98
DVT	3.43	4.02	4.56	6.02
Mortality	1.68	1.92	3.02	3.39
Readmission	6.01	6.32	6.92	6.88

Abbreviations: AA, ankle arthroplasty; DVT, deep vein thrombosis; PE, pulmonary embolism; ppt, percentage point; TAA, total ankle arthroplasty.

^aMDE = minimum detectable absolute difference (percentage points) for a 2-sample comparison of proportions with dementia vs non-dementia groups, assuming two-sided α = 0.05 and 80% power.