Anaesthetic choice for hip or knee arthroplasty in New Zealand: Risk of postoperative death and variations in use

Abstract

Anaesthetic choice for large joint surgery can impact postoperative outcomes, including mortality. The extent to which the impact of anaesthetic choice on postoperative mortality varies within patient populations and the extent to which anaesthetic choice is changing over time remain under-explored both internationally and in the diverse New Zealand context. In a national study of 199,211 hip and knee replacement procedures conducted between 2005 and 2017, we compared postoperative mortality among those receiving general, regional or general plus regional anaesthesia. Focusing on unilateral (n=86,467) and partial (n=13,889) hip replacements, we assessed whether some groups within the population are more likely to receive general, regional or general plus regional anaesthesia than others, and whether mortality risk varies depending on anaesthetic choice. We also examined temporal changes in anaesthetic choice over time. Those receiving regional alone or general plus regional for unilateral hip replacement appeared at increased risk of 30-day mortality compared to general anaesthesia alone, even after adjusting for differences in terms of age, ethnicity, deprivation, rurality, comorbidity, American Society of Anesthesiologists physical status score and admission type (e.g. general plus regional: adjusted hazard ratio (adj. HR)=1.94, 95% confidence intervals (CI) 1.32 to 2.84). By contrast, we observed lower 30-day mortality among those receiving regional anaesthesia alone compared to general alone for partial hip replacement (adj. HR=0.86, 95% CI 0.75 to 0.97). The latter observation contrasts with declining temporal trends in the use of regional anaesthesia alone for partial hip replacement procedures. However, we recognise that postoperative mortality is one perioperative factor that drives anaesthetic choice.

Keywords

Postoperative mortality perioperative mortality general anaesthesia

Introduction

There are a number of factors that influence the choice of anaesthetic for large joint surgeries, including hospital protocols, surgeon preferences, patient choice and patient factors such as age and comorbidity. With nearly 90% of all musculoskeletal procedures performed under general anaesthesia (see Supplemental Material S1), hip and knee arthroplasty are two of the relatively few procedures that are fairly interchangeably performed under regional/neuraxial (hereafter regional) anaesthesia.¹ Among those receiving regional compared to general anaesthesia, there is some evidence of reduced risk of postoperative infection,^2–⁵ reduced unplanned intubations,⁶ reduced risk of blood transfusion,^5,7 reduced rates of stroke, cardiovascular and other pulmonary complications,^2,6 reduced opioid consumption⁸ and reduced postoperative pain.⁸ Evidence from propensity score analyses, in which patients are matched based on a host of factors that might influence anaesthetic choice, suggests that there may be some postoperative mortality benefit to the use of regional anaesthesia over general anaesthesia in hip or knee arthroplasty,^9,10 although this has not been observed across all studies that have compared this factor between anaesthetic types.¹

The impact of anaesthetic type on risk of postoperative mortality for large joint surgeries remains unexplored in the New Zealand context, as does the extent to which anaesthetic choice varies by patient subgroup (age, ethnicity, comorbidity, etc.). Additionally, the extent to which the impact of anaesthetic type on risk of postoperative mortality varies by patient subgroup (e.g. age, comorbidity, ethnicity) also remains under-explored internationally.

In this article, we use comprehensive, high-quality national-level data to compare postoperative mortality outcomes among those receiving general versus regional anaesthesia in the context of hip and knee arthroplasty within New Zealand. First, we describe the main procedure types performed under each anaesthetic type, and then describe the age-adjusted rates of 30- and 90-day death for each procedure type. Next, using common procedures as exemplars, we assess (a) whether some groups within the population are more likely to receive regional anaesthesia than others, and (b) whether mortality risk varies within these specific strata depending on the choice of anaesthetic–adjusting for the impact of other factors that may contribute to both anaesthetic choice and risk of mortality following surgery (age, comorbidity, etc.).

Methods

Participants and data sources

Our study cohort included all individuals who underwent a hip or knee arthroplasty (International Classification of Diseases (ICD)-10-AM/ACHI (Australian Classification of Health Interventions) 3rd edition codes: hip arthroplasty 114,925; knee arthroplasty 84,286; see procedure codes in Supplemental Material S2) between 1 January 2005 and 31 December 2017. This cohort are a subset of a larger pan-procedure cohort which we have previously described,¹¹ with the methods of the current study overlapping with the methods used in that broader study. National-level hospitalisation data (National Minimum Dataset; NMDS¹²) were used to find eligible hospital admissions over the study period that included procedures. The NMDS is a record of all publicly funded (and most privately funded) inpatient events in New Zealand (a country with universal healthcare). Clinical coders code diagnoses and procedures at a regional level, and these data are then reported to the central NMDS collection.¹² To allow for patients who had more than one hip or knee arthroplasty during their admission, admissions were further divided into the individual days that procedures were performed during the admission. It was possible for individuals to have both a hip and a knee procedure during the study period, including during the same procedure–in which case, the two procedures were treated as separate events for the purposes of our analyses. We restricted our analysis to New Zealand residents to ensure follow-up for postoperative death.

Patient comorbidity at the start of each admission (see Variables below) was measured using data on recorded diagnoses from public and private hospital admission records for the five-year period prior to the admission. Date of death was defined using the date of death variable that appears on the National Health Index dataset.¹³

Variables

All variables were derived from the NMDS dataset. Primary anaesthetic type was defined using ICD codes (general anaesthesia 9251410–9251499; regional anaesthesia 9250610–9251299). Regional anaesthesia included both neuraxial block (9250610–9250899) and regional block (9250910–9251299). For a given procedure, primary anaesthesia was defined as that which occurred on the day of the surgical procedure. We created three separate anaesthetic groups: those who received general anaesthesia only, those who received regional anaesthesia only and those who received both a general and regional anaesthesia.

Patient age was defined at the time of procedure (categorised as < 55, 55–64, 65–74 or 75+ years). This age categorisation balanced clinical relevance against the size of the individual age groups, the latter of which will influence the statistical precision of our findings. Patient sex was defined at the time of the surgical procedure and categorised as either male or female, as only these two categories were recorded in the hospital data. Ethnicity data were sourced from NMDS (time of admission) and prioritised into mutually exclusive categories of Māori, Pacific, Asian, European or Middle Eastern/Latin American/African/other.¹⁴ NMDS ethnicity data collections are guided by ethnicity data protocols, which ideally require patient self-identification.

Area deprivation for each patient at the time of procedure was determined using New Zealand’s 2013 Deprivation Index (NZDep) for each individual’s residential address (and subsequent domicile code).¹⁵ NZDep deciles were collapsed into quintiles ranging from 1 (least deprived) to 5 (most deprived). NZDep data were missing for 1456 hip procedure patients (1% of cohort) and 1585 knee procedure patients (1% of cohort). Patient rurality was defined using a modified version of the Urban/Rural Profile Classification (URPC),¹⁶ with the area where the patient lived at the time of the procedure classified as urban (main urban area+satellite urban area), independent urban or rural. Rurality data were missing for 1080 hip procedure patients (1% of cohort) and 1187 knee procedure patients (1% of cohort).

Patient comorbidity was defined using the M3 Index.¹⁷ For each patient, NMDS data from the five-year period prior to the date of admission for the given procedure were searched for the presence of any of the 61 conditions included in the M3 Index using ICD-10 codes, and a final comorbidity score was calculated (see Stanley and Sarfati¹⁷ for weights and more information about the M3 Index). For descriptive analysis, M3 scores were categorised as 0, >0–1, >1–2, >2–3 or 3+, while for regression models, M3 score was included as a splined variable¹⁷ with knots placed at the 0th, 90th and 95th percentiles.¹⁸ American Society of Anesthesiologists (ASA) physical status score was determined from ICD anaesthesia codes and categorised as either 1–2, 3, 4–5 or unknown.¹⁹ In those instances where more than one ASA score was coded on a given procedure day, a give-way rule was applied wherein the worst ASA score was retained for that procedure day. All procedures were assigned an admission type, depending on the reason for their hospital admission, with admission type categorised as either acute or elective/waiting list. Postoperative mortality was defined as death from any cause recorded within either 30 or 90 days of that procedure, inclusive of the date of the procedure.

Statistical analysis

Descriptive analyses included crude counts and age-standardised rates (per 100 procedures). Because of the possibility that there might be differences in age distribution between those given general anaesthesia and those given regional anaesthesia, we age standardised rate data to the age structure of the entire combined arthroplasty population, in order to account for changes in population age structure over time.

In order to establish the independent relationship between anaesthetic type and risk of 30-day mortality within each of our patient subgroups (‘covariates’), we conducted Cox proportional hazards regression modelling. Within the Cox model, we censored procedures when either (a) no death occurred within 30 days of the procedure, or (b) if a prior procedure occurred within a 30-day period from the date of the original procedure.¹⁹ As such, for those who had multiple procedures within a given 30-day period, the first procedure was used as the index procedure. Hazard ratios (HR) and their 95% confidence intervals (CI) were determined for the risk of 30-day mortality between anaesthetic types for each modelled variable (age, sex, ethnicity, deprivation, rurality, M3 score, ASA score and admission type), giving the risk of mortality within each stratum of each covariate, adjusted for all other covariates in the model (with the exception of M3 and ASA, which were not mutually adjusted for each other due to possible collinearity).

Finally, we determined temporal trends in the crude proportion of procedures that were conducted under each of the anaesthesia types over the study period. To test for notable increases or decreases in the rate of a given anaesthesia type over time, we fitted a linear regression model to each procedure, separately for each anaesthetic group stratum, with procedure year as the independent variable and the number of procedures as the dependent variable. We output P-values from the resulting parameter estimates to augment our assessment of changes in anaesthetic use over time for each procedure. We repeated these analyses with further stratification by age group to assess whether changes over time were occurring consistently across age groups.

Data management and analysis was performed using SAS v9.4 (SAS Institute, Inc., Cary, NC, USA), Stata v16 (StataCorp LLC, College Station, TX, USA) and Microsoft Excel 2016 (Microsoft Corp., Redmond, WA, USA). The study sought and received ethical approval from the University of Otago Human Ethics Committee (reference: H18/085).

Results

The total number of hip or knee procedures and the deaths that occurred within 30 or 90 days of these procedures are presented in Table 1. The vast majority of the 114,925 hip procedures performed over the study period were unilateral hip replacements (72%), followed by hip revisions (14%), partial hip replacements (also known as hemiarthroplasty; 12%), bilateral hip replacements (1%) and hip excisions (1%). The vast majority of the 84,286 knee procedures were unilateral knee replacements (77%), followed by knee revisions (10%), partial knee replacements (9%) and bilateral knee replacements (4%).

Table 1.

Numbers of procedures and deaths by anaesthetic type, further stratified by procedure type.

	General anaesthesia only				Regional anaesthesia only				General plus regional anaesthesia
	Procedures	Deaths	Death rate ( n/100)		Procedures	Deaths	Death rate ( n/100)		Procedures	Deaths	Death rate ( n/100)
	n	n	Crude	Age adjusted	n	n	Crude	Age adjusted	n	n	Crude	Age adjusted
30-day mortality
Any hip	55,369	675	1.2	1.3 (1.2–1.4)	39,054	558	1.4	1.2 (1.1–1.3)	20,502	444	2.2	1.8 (1.6–2)
Unilateral hip replacement	42,197	62	0.1	0.2 (0.1–0.2)	31,768	86	0.3	0.3 (0.2–0.3)	12,502	58	0.5	0.5 (0.4–0.6)
Bilateral hip replacement	796	2	0.3	0.1 (0–1.2)	142	0	0	0 (0–3.4)	432	0	0	0 (0–1.9)
Partial hip replacement	5733	539	9.4	5.5 (4.7–6.4)	5071	443	8.7	7.8 (5.4–11)	3085	341	11.1	7.1 (5.7–8.9)
Hip excision	538	29	5.4	4.8 (3.2-6.9)	114	15	13.2	7.6 (4–15.3)	273	8	2.9	2.7 (1.1–5.5)
Hip revision	6105	43	0.7	0.6 (0.5–0.9)	1959	14	0.7	0.5 (0.3–0.9)	4210	37	0.9	0.7 (0.5–0.9)
Any knee	36,611	46	0.1	0.2 (0.1–0.2)	30,435	47	0.2	0.1 (0.1–0.2)	17,240	45	0.3	0.3 (0.2–0.4)
Unilateral knee replacement	28,110	26	0.1	0.1 (0.1–0.2)	27,010	37	0.1	0.1 (0.1–0.2)	13,458	31	0.2	0.2 (0.2–0.3)
Bilateral knee replacement	1125	0	0	0 (0–0.6)	335	1	0.3	1.1 (0–4.9)	882	2	0.2	0.2 (0–0.9)
Partial knee replacement	3925	2	0.1	0.1 (0–0.3)	1860	1	0.1	0 (0–0.3)	988	1	0.1	0.2 (0–0.7)
Knee revision	3451	18	0.5	0.9 (0.5–1.3)	1230	8	0.7	0.6 (0.2–1.2)	1912	11	0.6	0.6 (0.3–1)
90-day mortality
Any hip	55,369	1280	2.3	2.4 (2.2–2.5)	39,054	1122	2.9	2.3 (2.2–2.5)	20,502	768	3.7	3.1 (2.9–3.4)
Unilateral hip replacement	42,197	155	0.4	0.4 (0.4–0.5)	31,768	175	0.6	0.5 (0.5–0.6)	12,502	115	0.9	0.9 (0.8–1.1)
Bilateral hip replacement	796	2	0.3	0.1 (0–1.2)	142	0	0	0 (0–3.4)	432	0	0	0 (0–1.9)
Partial hip replacement	5733	1011	17.6	10.4 (9.3–11.6)	5071	901	17.8	13.9 (11.2–17.4)	3085	586	19	13.9 (11.5–16.8)
Hip excision	538	31	5.8	5.3 (3.6–7.6)	114	18	15.8	11.3 (5.4–21.7)	273	15	5.5	5.2 (2.8–8.7)
Hip revision	6105	81	1.3	1.2 (0.9–1.5)	1959	28	1.4	1.2 (0.8–1.8)	4210	52	1.2	1 (0.7–1.3)
Any knee	36,611	69	0.2	0.2 (0.2–0.3)	30,435	94	0.3	0.3 (0.2–0.4)	17,240	78	0.5	0.5 (0.4–0.6)
Unilateral knee replacement	28,110	42	0.1	0.2 (0.1–0.2)	27,010	74	0.3	0.3 (0.2–0.3)	13,458	48	0.4	0.4 (0.3–0.5)
Bilateral knee replacement	1125	0	0	0 (0–0.6)	335	1	0.3	1.1 (0–4.9)	882	4	0.5	0.4 (0.1–1.1)
Partial knee replacement	3925	4	0.1	0.2 (0.1–0.5)	1860	4	0.2	0.2 (0.1–0.6)	988	3	0.3	0.4 (0.1–1.2)
Knee revision	3451	23	0.7	1.1 (0.7–1.6)	1230	15	1.2	1.1 (0.6–1.8)	1912	23	1.2	1.2 (0.8–1.8)

Bracketed ranges are 95% confidence intervals.

With respect to rate of postoperative death, hip procedures could broadly be described as having either a lower rate of postoperative mortality (uni/bilateral hip replacement and hip revision; age-adjusted rates 0.1–0.7 deaths within 30 days/100 procedures) or higher rate (partial hip replacement and hip excision; 4–8 deaths within 30 days/100 procedures; Table 1). The pooled ‘any hip’ rate of 30- or 90-day mortality did not reflect the mortality profile of any of the individual procedures (1.2–1.8 deaths within 30 days/100 procedures), minimising its value as a representative measure of mortality risk following hip arthroplasty. All of the individual knee procedures carried a risk of postoperative death between 0.1 and 1.1 deaths within 30 days/100 procedures.

Due to low numbers of deaths for most individual procedures, it was difficult to establish patterns of differential mortality risk between anaesthetic type and arthroplasty procedure (Table 1). However, there was some evidence of differences between anaesthesia types for unilateral hip procedures (e.g. age-standardised rate of 30-day mortality: general 0.2/100 procedures, regional 0.3/100, general plus regional 0.5/100), as well as partial hip procedures (e.g. 30-day mortality: general 5.5/100 procedures, regional 7.8/100, general plus regional 7.1/100). There was a large shift in partial hip 30-day mortality rates once adjusted for age (and little shift for unilateral hip).

The lower mortality rates seen with bilateral hip replacements (age standardised rate of 30-day mortality: general anaesthesia 0.1/100 procedures, regional anaesthesia 0/100 procedures, general plus regional 0/100 procedures) compared to unilateral hip replacements (general 0.2/100, regional 0.3/100, general plus regional 0.5/100) likely reflects the relatively young healthy cohort that are offered bilateral hip replacements.

Because most procedures had a relatively low number of deaths, we restricted our subsequent covariate-stratified analyses to those procedures that were both commonly performed and had a high absolute number of deaths within 30 or 90 days. As such, subsequent reported results are limited to unilateral hip replacements (relatively low risk of mortality but common procedure) and partial hip replacements (less common procedure but relatively high risk of mortality).

Unilateral hip replacement

For all unilateral hip procedures, general anaesthesia was most common (49%), followed by regional anaesthesia (37%). Only 14% were done with a combination of general and regional anaesthesia (Table 2). There was some variation in anaesthetic type (general, regional or general plus regional) used for unilateral hip replacements across patient subgroups. The oldest age group (>75 years) was equally split across general and regional anaesthesia for their unilateral hip replacements (41% versus 43%). In comparison, those in the youngest cohort were more likely to have general anaesthesia than a regional anaesthesia (57% versus 27%). Māori and Pacific patients were more likely to receive regional anaesthesia than the European or Asian populations (age-adjusted proportion: Māori 54%, Pacific 50%, European 37%, Asian 33%). The use of regional anaesthesia also increased with increasing levels of deprivation (e.g. NZDep quintile 1 (least deprived): 27%; NZDep quintile 5: 49%). Those with ASA scores of 1–3 were more likely to have regional anaesthesia (52%–55%) than those with an ASA score of 4–5 (44%) or no recorded ASA score (12%). There was little difference in anaesthesia type by sex, comorbidity (M3 index) or acute/elective status.

Table 2.

Age-adjusted proportions of general or regional anaesthesia receipt, stratified by covariates and further stratified by procedure type.

	Unilateral Hip Replacement						Partial Hip Replacement
	General Only		Regional Only		General Plus Regional		General Only		Regional Only		General Plus Regional
	n	Row %^a	n	Row %^a	n	Row %^a	n	Row %^a	n	Row %^a	n	Row %^a
Total procedures	42,197	49%	31,768	37%	12,502	14%	5733	41%	5071	37%	3085	22%
Age (years)
<55	7027	57%	3356	27%	1970	16%	123	72%	16	9%	33	19%
55–64	11,529	54%	6861	32%	2775	13%	275	66%	70	17%	71	17%
65–74	13,779	48%	11,120	38%	4050	14%	621	49%	376	30%	263	21%
75+	9862	41%	10,431	43%	3707	15%	4714	39%	4609	38%	2718	23%
Sex
Female	22,039	47%	17,434	37%	7038	15%	3907	50%	3493	27%	2199	23%
Male	20,158	49%	14,334	37%	5464	14%	1826	55%	1578	27%	886	18%
Ethnicity
Māori	2131	28%	3537	54%	1348	18%	115	39%	121	30%	95	31%
Pacific	273	29%	393	50%	180	21%	55	30%	52	35%	47	36%
Asian	299	48%	196	33%	124	19%	99	48%	85	33%	83	19%
European	36,242	49%	26,593	37%	10,352	14%	5292	49%	4691	37%	2769	14%
MELAA/other	3252	67%	1049	23%	498	11%	172	67%	122	23%	91	11%
Deprivation (NZDep decile)
1–2	10,019	63%	4107	27%	1573	10%	996	63%	702	21%	442	15%
3–4	8402	52%	5357	35%	2072	13%	969	53%	953	28%	548	19%
5–6	9147	50%	6541	36%	2613	14%	1195	52%	1041	27%	715	21%
7–8	8771	45%	7566	40%	2899	15%	1498	52%	1319	29%	741	20%
9–10	5367	32%	7907	49%	3072	19%	999	45%	976	30%	597	26%
Rurality
Urban	28,657	50%	20,023	36%	8401	15%	4505	52%	3719	27%	2461	22%
Independent urban	5887	40%	6552	45%	2077	14%	783	50%	991	34%	423	16%
Rural	7103	50%	4868	37%	1737	13%	368	60%	278	22%	157	18%
Comorbidity (M3 index category)
0	31,514	51%	20,808	36%	7922	13%	1830	62%	1556	24%	793	15%
>0–1	8958	42%	8986	41%	3608	17%	1703	47%	1660	32%	991	22%
>1–2	1162	37%	1379	42%	660	21%	1618	45%	1317	29%	941	26%
>2–3	465	40%	475	39%	252	21%	438	42%	391	30%	269	28%
≥3	98	36%	120	41%	60	23%	144	32%	147	34%	91	34%
ASA score
1–2	8911	24%	19,410	55%	7661	21%	645	38%	864	36%	380	25%
3	3951	27%	8044	52%	3084	21%	2542	40%	2499	32%	1721	28%
4–5	229	32%	348	44%	175	24%	930	38%	905	34%	604	28%
Not recorded	29,106	83%	3966	12%	1582	5%	1616	73%	803	18%	380	9%
Admission type
Acute	1746	39%	1807	35%	1093	25%	5059	41%	4933	33%	2990	26%
Elective/waiting list	40,451	49%	29,961	37%	11,409	14%	674	78%	138	13%	95	9%

^aRow percentages (within procedure types) age-adjusted using direct age standardisation to the total arthroplasty population. Age-stratified data are unadjusted.

MELAA: Middle Eastern/Latin American/African; NZDep: New Zealand Deprivation Index; ASA: American Society of Anesthesiologists.

Once adjusted for age, sex, ethnicity, deprivation, rurality, comorbidity, ASA score and admission type, there was an increased risk of 30-day postoperative mortality following unilateral hip replacement among those receiving regional anaesthesia (adjusted (adj.) HR=1.35, 95% CI 0.95 to 1.92) or a combination of general plus regional (adj. HR=1.94, 95% CI 1.32 to 2.84) compared to those receiving general anaesthesia only (Table 3). Similar patterns of higher mortality from regional and even higher mortality for the combination of general and regional compared to general anaesthesia alone were seen across strata of age, sex and deprivation (Table 3). These differences in mortality risk attenuated somewhat for 90-day mortality (regional only: adj. HR=1.11, 95% CI 0.88 to 1.40; general plus regional: adj. HR=1.53, 95% CI 1.19 to 1.98; Supplemental Material S3).

Table 3.

Numbers of deaths and likelihood of 30-day mortality within patient subgroups, stratified by procedure type.

	Unilateral hip replacement						Partial hip replacement
	General only		Regional only		General plus regional		General only		Regional only		General plus regional
	Deaths (n)	Adj. HR (95% CI)	Deaths (n)	Adj. HR (95% CI)	Deaths (n)	Adj. HR (95% CI)	Deaths (n)	Adj. HR (95% CI)	Deaths (n)	Adj. HR (95% CI)	Deaths (n)	Adj. HR (95% CI)
Final model	62	Ref.	86	1.35 (0.95–1.92)	58	1.94 (1.32–2.84)	539	Ref.	443	0.86 (0.75–0.97)	341	1.09 (0.95–1.25)
Age (years)
<55	2	Ref.	4	–	1	–	4	Ref.	2	–	0	–
55–64	6	Ref.	5	1.14 (0.32–4.05)	7	2.58 (0.81–8.29)	5	Ref.	5	1.43 (0.37–5.54)	1	–
65–74	21	Ref.	13	0.48 (0.23–1.01)	17	1.38 (0.69–2.76)	24	Ref.	21	1.04 (0.57–1.89)	24	1.72 (0.96–3.09)
75+	33	Ref.	64	1.82 (1.17–2.84)	33	2.18 (1.31–3.62)	506	Ref.	415	0.84 (0.74–0.96)	316	1.08 (0.94–1.24)
Sex
Female	36	Ref.	42	1.17 (0.73–1.88)	38	2.16 (1.34–3.49)	324	Ref.	248	0.84 (0.71–0.99)	206	1.08 (0.9–1.28)
Male	26	Ref.	44	1.7 (1–2.89)	20	1.63 (0.87–3.08)	215	Ref.	195	0.88 (0.72–1.08)	135	1.11 (0.89–1.39)
Ethnicity
Māori	7	Ref.	2	–	2	-	7	Ref.	17	2.36 (0.93–5.99)	6	0.89 (0.29–2.7)
Pacific	1	Ref.	1	–	2	-	3	Ref.	3	–	3	–
Asian	1	Ref.	0	–	2	-	6	Ref.	5	0.89 (0.26–3.07)	13	2.32 (0.84–6.39)
European	53	Ref.	82	1.56 (1.08–2.27)	51	2.1 (1.4-3.15)	513	Ref.	412	0.85 (0.74–0.97)	311	1.09 (0.94–1.26)
MELAA/other	0	Ref.	1		1		10	Ref.	6	0.61 (0.2–1.81)	8	0.85 (0.31–2.32)
Deprivation (NZDep decile)
1–2	13	Ref.	13	1.58 (0.67–3.75)	9	2.28 (0.89-5.86)	76	Ref.	47	0.75 (0.52–1.08)	46	1.2 (0.83–1.74)
3–4	9	Ref.	14	1.46 (0.6–3.54)	8	1.93 (0.72-5.19)	80	Ref.	79	0.93 (0.68–1.28)	51	1.02 (0.71–1.45)
5–6	9	Ref.	19	2.04 (0.89–4.7)	13	2.84 (1.17-6.89)	115	Ref.	85	0.8 (0.6–1.06)	84	1.12 (0.84–1.49)
7–8	17	Ref.	24	1.5 (0.77–2.93)	17	2.48 (1.21-5.08)	150	Ref.	128	0.96 (0.76–1.22)	85	1.1 (0.84–1.45)
9–10	13	Ref.	16	0.7 (0.33–1.48)	10	1.08 (0.47-2.52)	112	Ref.	98	0.85 (0.65–1.12)	74	1.08 (0.8–1.46)
Rurality
Urban	43	Ref.	53	1.35 (0.88–2.08)	40	2 (1.27-3.16)	429	Ref.	327	0.86 (0.75–1)	269	1.07 (0.92–1.25)
Independent urban	14	Ref.	19	0.95 (0.46–1.95)	11	1.75 (0.78-3.96)	79	Ref.	87	0.84 (0.62–1.15)	55	1.23 (0.87–1.75)
Rural	4	Ref.	14	3.63 (1.08–12.24)	6	3.08 (0.77-12.27)	25	Ref.	23	0.82 (0.46–1.48)	16	0.95 (0.49–1.83)
Comorbidity (M3 index category)^a
0	22	Ref.	29	1.31 (0.75–2.3)	22	2.57 (1.41-4.68)	105	Ref.	85	0.71 (0.53–0.95)	69	1.22 (0.9–1.66)
>0-1	14	Ref.	40	2.68 (1.45–4.95)	16	2.33 (1.12-4.85)	136	Ref.	130	0.92 (0.72–1.17)	78	0.97 (0.74–1.29)
>1-2	11	Ref.	9	0.7 (0.28–1.74)	9	1.48 (0.59-3.7)	192	Ref.	133	0.81 (0.65–1.01)	117	1.07 (0.85–1.34)
>2-3	10	Ref.	6	0.67 (0.24–1.89)	10	1.75 (0.72-4.26)	72	Ref.	55	0.82 (0.57–1.17)	55	1.35 (0.94–1.93)
≥3	5	Ref.	2	–	1	-	34	Ref.	40	1.11 (0.69–1.77)	22	1.02 (0.59–1.77)
ASA score^b
1–2	10	Ref.	21	1.08 (0.51–2.32)	14	1.8 (0.8-4.07)	19	Ref.	28	1.05 (0.58–1.88)	14	1.31 (0.66–2.63)
3	19	Ref.	48	1.64 (0.95–2.82)	29	2.15 (1.2-3.85)	237	Ref.	170	0.71 (0.58–0.86)	161	1.02 (0.83–1.25)
4–5	10	Ref.	8	0.57 (0.22–1.48)	9	1.07 (0.42-2.73)	180	Ref.	178	0.99 (0.8–1.22)	129	1.13 (0.9–1.42)
Not recorded	23	Ref.	9	1.25 (0.55–2.88)	6	2.01 (0.76-5.31)	103	Ref.	67	0.88 (0.65–1.21)	37	1.14 (0.78–1.66)
Admission type
Acute	33	Ref.	38	1.12 (0.7–1.8)	28	1.34 (0.8-2.24)	525	Ref.	431	0.84 (0.74–0.96)	331	1.07 (0.93–1.23)
Elective/waiting list	29	Ref.	48	1.51 (0.89–2.57)	30	2.67 (1.5-4.75)	14	Ref.	12	1.4 (0.61–3.23)	10	1.78 (0.75–4.22)

^aModel excludes ASA score.

^bModel excludes comorbidity. Hazard ratios are not shown where fewer than five deaths occurred over the follow-up period. Adj. HR: adjusted hazard ratio; CI: confidence intervals; MELAA: Middle Eastern/Latin American/African; NZDep: New Zealand Deprivation Index; ASA score: American Society of Anesthesiologists physical status score.

Partial hip replacement

Overall, partial hip replacements were most commonly performed under general or regional anaesthesia (41% versus 37%) and less commonly with a combination of general and regional anaesthesia (22%). The majority of partial hip replacements occurred in those aged >75 years of age, and for this age group, these were equally likely to be completed under general or regional anaesthesia (39% versus 38%) and less likely to be completed under the combination of general and regional (23%). In contrast, the small number of partial hip replacements occurring in the youngest cohort were largely done under general anaesthesia (77%), with only 9% under regional and the remaining 19% under a combination of general and regional. Māori and Pacific patients were less likely to have general anaesthesia only and more likely to have general plus regional than their European counterparts (e.g. age-adjusted proportions of general plus regional: Māori 31%, Pacific 36%, European 14%). Like unilateral hip replacement, the rate of regional only and general plus regional increased with increasing levels of area deprivation. The proportion of patients receiving general plus regional anaesthesia increased markedly with increasing comorbidity, with a corresponding reduction in general anaesthesia alone. Those admitted acutely for partial hip replacement were far more likely to have a regional only (33%) or general plus regional (26%) compared to elective/waiting list procedures (regional 13%, general plus regional 9%).

In contrast to the findings for unilateral hip replacements, overall, those receiving regional anaesthesia for partial hip replacements were around 15% less likely to die than those receiving general anaesthesia alone, after adjusting for all measured covariates (adj. HR=0.86, 95% CI 0.75 to 0.97; Table 3). There was little difference in the mortality risk between those receiving general anaesthesia alone and general plus regional anaesthesia for their partial hip replacements once adjusted for measured covariates (adj. HR=1.09, 95% CI 0.95 to 1.25). In stark contrast to the overall pattern of reduced mortality risk for regional anaesthesia for partial hip replacement, Māori had two and a half times the risk of 30-day postoperative mortality from partial hip surgery from regional anaesthesia compared to general anaesthesia after adjusting for the other measured confounders (adj. HR=2.51, 95% CI 1.16 to 5.43).

Temporal trends in anaesthetic use

Temporal trends in the use of the three anaesthetic types are shown in Figure 1 for both unilateral and partial hip replacements. For unilateral hip replacements, there was an increase in the use of both general anaesthesia alone and regional anaesthesia alone over the study period, with a corresponding substantial reduction in the use of general plus regional anaesthesia (P < 0.0001). This pattern occurred consistently across all age groups (P < 0.001; data not shown). For partial hip replacement, there was a reduction in the use of regional anaesthesia alone over time (P < 0.001), and a corresponding increase in both general anaesthesia alone (P < 0.05) and general plus regional anaesthesia (P < 0.001).

Figure 1.

Temporal trends in anaesthetic use over time for unilateral (top) and partial (bottom) hip replacement.

Discussion

Using national-level data in the context of universal healthcare, we found that over the 2005–2017 period, those unilateral hip replacement patients who received regional anaesthesia alone or general plus regional appeared at increased risk of 30-day mortality compared to those receiving general anaesthesia alone. In contrast, there was lower 30-day mortality among those receiving regional anaesthesia alone compared to general anaesthesia alone for partial hip replacements.

Our finding of greater risk of 30-day postoperative mortality for unilateral hip replacements under regional anaesthesia alone (adj. HR=1.35, 95% CI 0.95 to 1.92) and for those receiving a combination of general and regional anaesthesia (adj. HR=1.94, 95% CI 1.32 to 2.84) compared to general anaesthesia alone is in contrast to the findings of a previous US study of 180,000 total hip replacements by Memtsoudis et al.¹⁰ After adjusting for similar factors to those that we adjusted for (age, sex, ethnicity, deprivation, rurality, comorbidity, ASA score and admission type), Memtsoudis et al. found evidence of elevated mortality risk among those receiving general anaesthesia alone (general versus regional (reference): adjusted odds ratio (OR)=1.28, 95% CI 0.70 to 2.37; general versus general plus regional: OR=1.55, 95% CI 0.88 to 2.70). We remain uncertain why this difference may exist, although it may be related to differences in the populations under study (e.g. the Memtsoudis study included patients from 400 select hospitals across the USA,¹⁰ and so findings may not be generalisable to whole-population studies such as the current one). There may have been other (unmeasured) differences between the studies (including both populations and study design) that reduce the comparability of our results.

In contrast to the reduced likelihood of mortality among those receiving general anaesthesia alone for unilateral hip replacements, the evidence for partial hip replacement is mixed. We found a 14% mortality benefit for the use of regional anaesthesia alone versus general anaesthesia alone for partial hip replacements (adj. HR=0.86, 95% CI 0.75 to 0.97). This pattern was observed across most patient subgroups. A crucial exception to this was observed for Māori patients, who appeared more than twice as likely to die following regional compared to general anaesthesia (adj. HR=2.36, 95% CI 0.93 to 5.99). We note that the precision of these results will be substantially impacted by the low absolute numbers of Māori deaths following partial hip replacement (seven for general anaesthesia alone; 17 for regional anaesthesia alone). They are a further suggestion that the Māori experience of postoperative outcomes contrasts to that encountered by other New Zealanders. By way of background, we know that Māori have higher postoperative mortality following hip arthroplasty than non-Māori (adjusted OR=2.60, 95% CI 1.20 to 5.62),²⁰ with this higher postoperative mortality occurring across many other procedure types, including acute and elective coronary artery bypass graft²¹ and acute and elective colorectal procedures.^20,22 The differential mortality experience of Māori receiving regional anaesthesia alone (compared to Māori receiving general anaesthesia alone) may indicate that the factors that could impart a protective effect among those receiving regional anaesthesia–such as reduced risk of postoperative infection,²^–5 unplanned intubations,⁶ blood transfusion,^5,7 stroke, cardiovascular and other pulmonary complications^2,6–are not equally shared by Māori receiving regional anaesthesia. There may also be differential perioperative management of comorbid conditions among those Māori receiving regional anaesthesia (compared to those receiving general), as well as differential surgeon or anaesthetist experience and the overall quality of perioperative care received among these patients. These factors require further focused investigation.

We note that our overall observations of an apparent association with lower mortality with regional anaesthesia alone contrast with two recent systematic reviews and meta-analyses of hip fracture procedures, where the authors found either no difference between anaesthesia type in terms of 30-day mortality²³ or a marginal mortality benefit to general anaesthesia (pooled OR=1.15, 95% CI 1.01 to 1.32).²⁴ However, the authors of the latter review warned that the included studies had a high risk of confounding. This is relevant when we consider that, in our own study, 30-day mortality was higher for regional (7.8/100) compared to general anaesthesia (5.5/100; crude rate ratio 1.4) before we adjusted for the confounding impact of our included covariates. The two randomised controlled trials that were included in their review²⁴ found a similar protective effect of regional anaesthesia as found in our study, although statistical precision was impacted by small sample sizes (pooled OR=0.74, 95% CI 0.26 to 2.14). We note that a larger randomised controlled trial entitled REGAIN (REgional versus General Anesthesia for promoting INdependence after hip fracture) is currently underway,²⁵ which has mortality as a secondary outcome and should be reporting in late 2021.

Temporal trends in anaesthetic use

For unilateral hip replacement, the use of general anaesthesia alone and regional anaesthesia alone increased substantially over the study period, while the use of general plus regional anaesthesia dropped. In light of the apparent increased mortality risk of general plus regional anaesthesia compared to general anaesthesia, this temporal shift appears in keeping with best practice. By contrast, the apparent reduced likelihood of mortality with regional anaesthesia alone for partial hip replacement is at odds with the clear and substantial temporal shift towards increased use of general anaesthesia for this procedure over our study period (either alone or in combination with regional anaesthesia). Since we have adjusted for age and comorbidity within our analysis, this temporal shift should not be confounded by population ageing and increased comorbid burden over time (although residual confounding is still feasible, as noted below). This temporal change towards general anaesthesia has been observed for other hip procedures and may reflect improvements in the availability of short-acting intravenous anaesthetics and laryngeal mask airway technology, and the increased use of peripheral nerve blocks.²⁶ In addition, the use of regional anaesthesia fixes the time available to complete the procedure from the moment it is administered, whereas this factor can be modulated when using general anaesthesia. From a morbidity perspective, this temporal change may also reflect the movement towards earlier mobilisation as part of Enhanced Recovery After Surgery (ERAS) protocols. Collectively, these factors might help to explain why we have observed a gradual movement towards general anaesthesia over time. However, the contrast between temporal trends in regional anaesthesia use for partial hip replacement and the apparent reduced likelihood of mortality for patients receiving this anaesthetic type require us to consider whether continuing this temporal trend is in the best interests of the patient. As recognised above, postoperative mortality is one perioperative factor that drives anaesthetic choice, and the protocols that govern anaesthetic choice should prioritise patient outcomes (including mortality and morbidity outcomes).

Strengths and limitations

The current study used high-quality national-level data across a relatively wide time interval, allowing us to look at both overall and temporal trends. We acknowledge that the current study does not capture within- or between-practice variation in anaesthesia use. Rather, it presents a national ‘average’. A more nuanced investigation of practice variation, and how this has changed over time, is of interest but was beyond the scope of the current study. We also acknowledge that there are other limitations to the data used for this study–for example, this study does not discern whether a given individual died of factors directly relating to their procedure, their underlying comorbidities or other reasons.

While our study includes data for a number of important confounding factors, including level of comorbidity, it is possible that we have not adequately controlled for potentially important differences between anaesthesia groups in terms of factors that might independently influence mortality risk. Factors such as surgeon and anaesthetist experience and techniques, as well as preoperative optimisation and the quality of postoperative care (including perioperative blood pressure management),²⁷ could vary between groups, and this will not have been captured within our study.

Conclusions

In a national study of nearly 200,000 hip and knee replacement procedures conducted between 2005 and 2017, we found some key differences in the risk of postoperative mortality between those receiving general anaesthesia only, regional anaesthesia only and general plus regional anaesthesia. We found that those receiving regional alone or general plus regional for unilateral hip replacement appeared at increased risk of 30-day mortality compared to those receiving general anaesthesia, even after adjusting for differences between anaesthesia groups in terms of age, ethnicity, deprivation, rurality, comorbidity, ASA score and admission type (e.g. general plus regional: adj. HR=1.94, 95% CI 1.32 to 2.84). By contrast, we observed lower 30-day mortality among those receiving regional anaesthesia alone compared to general alone for partial hip replacements (adjusted HR=0.86, 95% CI 0.75 to 0.97). This finding is at odds with declining temporal trends in the use of regional anaesthesia alone for partial hip replacement procedures. However, we recognise that postoperative mortality is one perioperative factor that drives anaesthetic choice. Protocols that govern anaesthetic choice should prioritise patient outcomes, including mortality and morbidity.

Footnotes

Author Contribution(s)

Jason K Gurney: Conceptualization; Data curation; Formal analysis; Funding acquisition; Investigation; Methodology; Project administration; Writing – original draft; Writing – review & editing.

Melissa A McLeod: Conceptualization; Investigation; Methodology; Writing – review & editing.

Douglas Campbell: Investigation; Methodology; Writing – review & editing.

Elizabeth Dennett: Investigation; Writing – review & editing.

Sarah Jackson: Investigation; Writing – review & editing.

Jonathan Koea: Investigation; Methodology; Writing – review & editing.

Nicholas Lash: Investigation; Methodology; Writing – review & editing.

Dick Ongley: Investigation; Methodology; Writing – review & editing.

Acknowledgements

We would like to acknowledge Chris Lewis from the Ministry of Health’s National Collections team for assistance with data extraction.

Declaration of conflicting interests

The authors declare no competing interests.

Funding

The authors disclosed receipt of the following financial support for the research, authorship and/or publication of this article: This study was funded by the Health Research Council of New Zealand (HRC reference # 18/037).

ORCID iD

Jason K Gurney

Supplemental material

Supplemental material for this article is available online.

References

Johnson

Kopp

Burkle

, et al. Neuraxial vs general anaesthesia for total hip and total knee arthroplasty: a systematic review of comparative-effectiveness research. Br J Anaesth 2016; 116: 163–176.

Helwani

Avidan

Ben Abdallah

, et al. Effects of regional versus general anesthesia on outcomes after total hip arthroplasty: a retrospective propensity-matched cohort study. J Bone Joint Surg Am 2015; 97: 186–193.

Liu

Elkassabany

, et al. Neuraxial anesthesia decreases postoperative systemic infection risk compared with general anesthesia in knee arthroplasty. Anesth Analg 2013; 117: 1010–1016.

Scholten

Leijtens

Hannink

, et al. General anesthesia might be associated with early periprosthetic joint infection: an observational study of 3,909 arthroplasties. Acta Orthop 2019; 90: 554–558.

Wilson

Farley

Erens

, et al. General vs spinal anesthesia for revision total knee arthroplasty: do complication rates differ? J Arthroplasty 2019; 34: 1417–1422.

Basques

Toy

Bohl

, et al. General compared with spinal anesthesia for total hip arthroplasty. J Bone Joint Surg Am 2015; 97: 455–461.

Walker

Nguyen

Schmidt

, et al. Postoperative outcomes associated with neuraxial vs general anesthesia following bilateral total knee arthroplasty. J Arthroplasty 2017; 32: 3632–3636.

Donauer

Bomberg

Wagenpfeil

, et al. Regional vs. general anesthesia for total knee and hip replacement: an analysis of postoperative pain perception from the international PAIN OUT registry. Pain Pract 2018; 18: 1036–1047.

Hutton

Brull

Macfarlane

AJR.

Regional anaesthesia and outcomes. BJA Educ 2018; 18: 52–56.

10.

Memtsoudis

Sun

Chiu

, et al. Perioperative comparative effectiveness of anesthetic technique in orthopedic patients. Anesthesiology 2013; 118: 1046–1058.

11.

Gurney

McLeod

Stanley

, et al. Postoperative mortality in New Zealand following general anaesthetic: demographic patterns and temporal trends. BMJ Open 2020; 10: e036451.

12.

12. Ministry of Health. National minimum dataset data mart dictionary. Wellington: Ministry of Health, 2020.

13.

13. Ministry of Health. National health index data dictionary. Wellington: Ministry of Health, 2009.

14.

Health Information Standards Organisation. HISO 10001:2017 ethnicity data protocols. Wellington: Ministry of Health, 2017.

15.

Salmond

Crampton

Development of New Zealand’s Deprivation Index (NZDep) and its uptake as a national policy tool. Can J Public Health 2012; 103: S7–S11.

16.

Robson

Purdie

Cormack

Unequal Impact II: Māori and non-Māori cancer statistics by deprivation and rural–urban status, 2002–2006. Wellington: Ministry of Health, 2010.

17.

Stanley

Sarfati

Development and validation of the M3 index for measuring multimorbidity load for risk of mortality. J Clin Epidemiol 2017; 95: 99–110.

18.

Harrell

Jr.

Regression modeling strategies: with applications to linear models, logistic regression, and survival analysis. 2nd ed. Cham: Springer, 2015.

19.

Perioperative Mortality Review Committee. Perioperative mortality in New Zealand: fifth report of the Perioperative Mortality Review Committee. Wellington: Health Quality and Safety Commission, 2016.

20.

Perioperative Mortality Review Committee. Perioperative mortality in New Zealand: inaugural report of the Perioperative Mortality Committee. Wellington: Health Quality and Safety Commission, 2011.

21.

Perioperative Mortality Review Committee. Perioperative mortality in New Zealand: fourth report of the Perioperative Mortality Review Committee. Wellington: Health Quality and Safety Commission, 2015.

22.

Perioperative Mortality Review Committee. Perioperative mortality in New Zealand: third report of the Perioperative Mortality Review Committee. Wellington: Health Quality and Safety Commission, 2014.

23.

Van Waesberghe

Stevanovic

Rossaint

, et al. General vs. neuraxial anaesthesia in hip fracture patients: a systematic review and meta-analysis. BMC Anesthesiol 2017; 17: 87.

24.

O’Donnell

McLoughlin

Patterson

, et al. Perioperative outcomes in the context of mode of anaesthesia for patients undergoing hip fracture surgery: systematic review and meta-analysis. Br J Anaesth 2018; 120: 37–50.

25.

ClinicalTrials.gov. REgional versus General Anesthesia for promoting INdependence after hip fracture (REGAIN): the University of Pennsylvania, https://clinicaltrials.gov/ct2/show/NCT02507505 (accessed 9 June 2020).

26.

Cozowicz

Poeran

Memtsoudis

SG.

Epidemiology, trends, and disparities in regional anaesthesia for orthopaedic surgery. Br J Anaesth 2015; 115: 57–67.

27.

Siletz

Childers

Faltermeier

, et al. Surgical technical evidence review of hip fracture surgery conducted for the AHRQ safety program for improving surgical care and recovery. Geriatr Orthop Surg Rehabil 2018; 9: 2151459318769215.