Is there a reduction in risk of revision when 36-mm heads instead of 32 mm are used in total hip arthroplasty for patients with proximal femur fractures?

Figures & data

Figure 1. Use of 32- and 36-mm heads in MoXLPE and CoXLPE THA after proximal femur fracture in NARA database.

Figure 2. Flowchart of the selection and matching process.

Figure 4. A directed acyclic graph (DAG) was constructed under the following assumptions:

1) THA ‘revision’ is dependent on ‘head size’, ‘patient age’, ‘sex’, ‘year of surgery’, surgical ‘approach’, and type of THA ‘fixation’. Choice of ‘head material’ is not expected to affect ‘revision’ due to the short follow-up of the study.

2) Choice of ‘head size’ is dependent on ‘approach’, ‘year of surgery’, ‘sex’, and ‘patient age’ as surgeons operating on older patients through a posterior approach have presumably chosen a larger head in order to, hopefully, reduce the risk of dislocation. Male patients, operated more recently, have probably received a larger head due to their larger acetabulum and because the use of larger heads has become more popular with time.

3) ‘Fixation’ is dependent on ‘year of surgery’ and ‘age’ because patients operated more recently have probably received an uncemented THA, due to the popularization of this technique, and older patients have probably received a cemented THA due to their poorer bone quality.

4) ‘Head material’ is dependent on ‘head size’ and ‘patient age’ because surgeons have probably chosen ceramic over metal heads in younger patients and when choosing larger heads due to the presumed lower polyethylene wear.

Provided that our assumptions are correct, adjusting for ‘patient age’, ‘sex’, ‘year of surgery’, and ‘approach’ in the multivariable Cox regression model should block all backdoor pathways (for variables available in our database) confounding the effect of ‘head size’ on ‘revision’.

Table 2. Descriptive statistics of study population after propensity score matching. Values number (%) unless otherwise specified

	32-mm head	36-mm head
	(n = 2,515)	(n = 2,515)	ASDM ^a
Follow-up, years ^b	2.4 (0.9–4.4)	2.6 (1.1–4.3)	0.03
Mortality	477 (19)	507 (20)	0.03
Age (standard deviation)	70 (11)	71 (11)	0.07
Year of surgery ^b	2013	2013	0.05
	(2011–2015)	(2011–2015)
Female sex	1,570 (62)	1,453 (58)	0.09
Cemented THA	823 (33)	813 (32)	0.02
Cementless THA	1,148 (46)	1,068 (43)	0.06
Hybrid THA	428 (17)	542 (22)	0.1
Reverse hybrid	116 (5)	92 (4)	0.06
MoXLPE ^c	2,108 (84)	2,152 (86)	0.06
CoXLPE ^d	407 (16)	363 (14)	0.05
Posterior approach	1724 (69)	1705 (68)	0.02

^a–dSee Table 1.

Table 1. Descriptive statistics of study population before propensity score matching. Values number (%) unless otherwise specified

	32-mm head	36-mm head
	(n = 8,957)	(n = 3,519)	ASDM ^a
Follow-up, years ^b	2.4 (1.0–4.4)	2.3 (1.0–4.0)	0.1
Mortality	1,702 (19)	621 (18)	0.04
Age (standard deviation)	73 (10)	70 (11)	0.2
Year of surgery ^b	2013	2014	0.1
	(2011–2015)	(2012–2015)
Female sex	6,266 (70)	1,812 (52)	0.4
Cemented THA	6,276 (70)	813 (23)	0.6
Cementless THA	1,219 (14)	1,729 (50)	0.7
Hybrid THA	430 (5)	885 (25)	0.5
Reverse hybrid	1,032 (12)	92 (3)	0.6
MoXLPE ^c	7,954 (89)	3,083 (88)	0.04
CoXLPE ^d	1,003 (11)	436 (12)	0.04
Posterior approach	3,912 (44)	2,599 (74)	0.7

^aAbsolute standardized difference in means.

^b Median (interquartile range).

^c Metal head on cross-linked polyethylene.

^d Ceramic head on cross-linked polyethylene.

Figure 5. Kaplan–Meier survival function for THA with 32- and 36-mm heads with endpoint revision for any reason (left panel) and revision due to dislocation (right panel).

Table 4. Cox proportional hazards models with endpoint revision for any reason and due to dislocation

Outcome Head size	Univariable model HR (CI)	Multivariable model HR (CI)
Revision for any reason
32-mm	1	1
36-mm	0.9 (0.7–1.2)	0.9 (0.7–1.2)
Revision due to dislocation
32-mm	1	1
36-mm	0.8 (0.5–1.3)	0.8 (0.5–1.3)

The multivariable model was adjusted for patient age, sex, year of surgery, and type of surgical approach.

HR (CI) = Hazard ratio (95% confidence interval)

Table 3. Kaplan–Meier survival estimates (%) at 1, 3, and 7 years for 32- and 36-mm heads with endpoint revision for any reason

Follow-up	Patients at risk	Cumulative revisions	Cumulative survival rate (CI)
32-mm heads
1-year	1,816	88	95.8 (95.1–96.7)
3-year	1,032	108	94.4 (93.4–95.5)
7-year	223	119	92.8 (91.2–94.4)
36-mm heads
1-year	1,922	87	95.9 (95.1–96.7)
3-year	1,099	102	95.2 (94.3–96.1)
7-year	140	111	93.7 (92.2–95.2)

CI = 95% confidence intervals.

Table 5. Kaplan–Meier survival estimates (%) at 1, 3, and 7 years for 32- and 36-mm heads with endpoint revision due to dislocation

Follow-up	Patients at risk	Cumulative revisions	Cumulative survival rate (CI)
32-mm heads
1-year	1,816	33	98.4 (97.9–98.9)
3-year	1,032	39	98.1 (97.6–98.7)
7-year	223	40	97.8 (97.0–98.7)
36-mm heads
1-year	1,922	28	98.7 (98.2–99.2)
3-year	1,099	31	98.6 (98.2–99.1)
7-year	140	33	98.3 (97.6–99.0)

CI = 95% confidence intervals.