Improving 1-Year Outcomes of Infrainguinal Limb Revascularization

Introduction

Mortality and amputation rates among patients undergoing endovascular or surgical revascularization for lower limb peripheral artery disease (PAD) have been examined in randomized controlled trials and observational studies based on routinely collected hospitalization data from the United States and Western European countries.^1–7 Over the past decades, both types of studies have found high rates of both death and major lower limb amputation among patients undergoing revascularization in the United Kingdom.^1,5,6 For example, in a study based on hospital admission records, the investigators reported a 1-year amputation rate of 9.2% and a 1-year mortality rate of 16.1% after femoropoliteal bypass surgery performed in England in 2006.⁶ This has been a concern, in particular, in comparison with corresponding rates from other countries, such as Germany, Finland, and different parts of the United States.^2–4,7

The availability of revascularization procedures has changed during the past decade (2006–2015), and with recent developments in endovascular and surgical technology, in particular, stents and drug-eluting technologies, less invasive procedures have become more widely used in the United Kingdom and elsewhere.^8,9 However, the impact of the increased availability and diversification of procedures on patient outcomes, such as major lower limb amputation and death, is unclear. For many patients, both surgical and endovascular procedures are deemed suitable alternative revascularization strategies, and randomized controlled trials comparing the outcomes of these procedures have aimed to produce information to guide clinical decision making where equipoise exists.^10,11 In many cases, however, patient fitness or anatomy dictate that one treatment modality is preferable to the other. Patient selection will thus influence the patient outcomes after revascularization procedures, which needs to be taken into account in analyses of routinely collected data.

To investigate whether developments in care have translated to improvements in patient outcomes, we examined the 1-year risks of major lower limb amputation and death following endovascular and surgical lower limb revascularization for infrainguinal PAD undertaken between 2006 and 2015 in England. Our analyses were based on individual-patient records from a nationwide set of routinely collected hospital admissions data, Hospital Episode Statistics (HES).

Methods

Data Sources

This study is exempt from United Kingdom National Research Ethics Committee approval because it involved secondary analysis of an existing data set of anonymized data. We do not have permission to share patient-level HES data and are therefore unable to make the data or study materials available to other researchers for replication purposes. HES data are available from the NHS Digital Data Access Advisory Group ([email protected]) for studies that meet the criteria for access to confidential data. HES data were made available by the NHS Digital (© 2015, reused with the permission of NHS Digital; all rights reserved).

Individual-level data on endovascular and surgical lower limb revascularizations performed between January 1, 2003, and December 31, 2015, were obtained from HES. This nationwide administrative data set captures information on all hospital admissions in National Health Service (NHS) hospitals in England.¹² Information on lower limb amputations following revascularization was obtained from HES, and mortality was ascertained by linking the patients’ HES records (using encrypted individual identifiers) to Office for National Statistics records of deaths registered in England up to the end of December 2015.¹³

Study Population

Our study population was men and women aged ≥35 years who underwent their first lower limb revascularization for infrainguinal PAD (index procedure) during the 10-year period from January 2006 to December 2015. Patients with a HES record of a revascularization up to 3 years before the index procedure (back to 2003) were excluded. We also excluded patients undergoing iliac procedures and those having revascularization because of cancer or trauma. Patient records with missing data on covariates (patient demographics, indications for revascularization, or comorbidities) were also excluded (<1% of all potentially eligible patients).

Procedures and Patient Outcomes

The revascularization procedures were divided into 2 groups: endovascular (angioplasty as the only revascularization, with or without stent) and surgical revascularizations (endarterectomy or profundaplasty as the only revascularization, or leg bypass either as the only revascularization or in combination with other revascularization procedures). The Office for Population Censuses and Surveys version 4 codes used to identify these procedures are provided in Tables I and II in the online-only Data Supplement.

The primary outcomes in our analyses were any major lower limb amputation and death from any cause occurring within 1 year of revascularization. The Office for Population Censuses and Surveys codes for identifying major lower limb amputations are provided in Table III in the online-only Data Supplement.

Patient Characteristics

HES data included information on patient age, sex, comorbidities, and indications for revascularization. Indications and comorbidities were ascertained from the International Classification of Diseases, 10th Revision diagnostic codes recorded at the relevant admission. Indications for revascularization were identified from the diagnostic codes recorded at the index admission and defined as follows: IC, intermittent claudication; SLI1, severe limb ischemia without a record of tissue loss; SLI2, severe limb ischemia with a record of ulceration; and SLI3, severe limb ischemia with a record of gangrene or osteomyelitis. Severe limb ischemia was defined as PAD or diabetes mellitus with peripheral circulatory complications (Table IV in the online-only Data Supplement). Comorbidities were coded into the Royal College of Surgeons Charlson Score, as previously reported.¹⁴ In brief, the score was defined as the number of selected comorbidities recorded at the index admission (revascularization), with the exception of acute conditions (such as myocardial infarction), which were counted as comorbidities if present in a HES record of a hospital admission within the 12 months before the index procedure (Table V in the online-only Data Supplement). PAD was used to define the indication for intervention and thus not included in the Royal College of Surgeons Charlson Score. Comorbid diabetes mellitus was defined as a record of diabetes mellitus in the index admission or up to a year before it (Table VI in the online-only Data Supplement).

Statistical Analyses

The risks of major amputation and death were examined by using Fine-Gray competing risks regression.^15,16 We examined unadjusted and multivariable-adjusted associations of each procedure with each outcome (amputation and death), with the other outcome as a competing risk. Analyses were stratified by diabetes status. Linear, logistic, and competing risks regression models were used to examine trends across the study period. The time to event was defined as months from the first revascularization procedure to major lower limb amputation, death of the patient, or the end of follow-up (1 year after revascularization). In the competing risk models, the proportionality of subdistribution hazards was checked by including an interaction term with time in the model. The assumption was valid for all procedure-outcome pairs. Age (10-year bands), sex, the Royal College of Surgeons Charlson Score (0, 1, 2, 3+), and indications for revascularization (IC, SLI1, SLI2, and SLI3) were analyzed as categorical variables and comorbid diabetes mellitus as a binary variable. The adjusted cumulative incidence rates are shown for the specified values of the main exposure, at average values of the covariates. All analyses were conducted using Stata MP 14 (Stata Corp).

Results

The characteristics of the patients included in our analyses are summarized in the Table. In total, 77 213 men and women underwent endovascular revascularization and 26 721 underwent surgical revascularization for infrainguinal PAD between January 2006 and December 2015. The median follow-up was 12 months (range, 5 days to 12 months). Most patients were men, and the majority were aged ≥65 years. The age distribution remained similar throughout the study period, whereas the proportion of men among patients undergoing revascularization increased slightly from 62.6% to 64.8%. There was also an increase in the proportions of patients with Royal College of Surgeons Charlson Scores 2 and 3, and in the proportions of patients undergoing revascularization for more severe limb ischemia (SLI1, SLI2, and SLI3). The prevalence of diabetes mellitus in the study population increased by ≈8% over the study period. Overall, endovascular procedures became more common and surgical revascularization less common over the study period. The proportions of patients who underwent an amputation or died during the year following revascularization decreased for both endovascular and surgical procedures (Table).

Adjusted cumulative incidences of major lower limb amputation and death following endovascular and surgical revascularization are shown in Figures 1 and 2, separately for each 2-year interval. In the beginning of our study period, 2006 to 2007, the 1-year cumulative incidence of amputation was 5.7% after endovascular procedures and 11.2% after surgical revascularization (Figure 1). By 2014 to 2015, the risk of amputation following endovascular procedures had decreased to 3.9% (P<0.0001), and the same risk after surgery had decreased to 6.6% (P<0.0001). (Unadjusted estimates are provided in Figures I and II in the online-only Data Supplement and Table VII in the online-only Data Supplement.)

• Download figure
• Download PowerPoint

The adjusted 1-year cumulative incidence of death at the beginning of our study period, 2006 to 2007, was 9.5% following endovascular procedures and 11.1% following surgery. Both decreased during the following biyearly intervals, with the cumulative incidence of death in 2014 to 2015 falling to 6.0% after endovascular and 6.4% after surgical revascularization (P<0.0001) (Figure 2).

• Download figure
• Download PowerPoint

Cumulative incidences of major amputation and death within 1 year of revascularization, stratified by procedure and diabetes status, are shown in Figures 3 and 4. (The corresponding unadjusted estimates are provided in the Appendix in the online-only Data Supplement and Tables VIII and IX in the online-only Data Supplement.) The cumulative incidence of each outcome was higher among diabetic patients than among patients with no record of diabetes mellitus, in both procedure groups and throughout the study period (Figures 3 and 4).

• Download figure
• Download PowerPoint

• Download figure
• Download PowerPoint

Figure 5 shows the secular change in the risks of amputation and death by procedure and diabetes status. In addition to the risk estimates, subdistribution hazard ratios are presented: they indicate the change in the cumulative incidence functions for amputation and death during the biyearly intervals of follow-up in comparison with the baseline 2006 to 2007. The 1-year cumulative incidence of major amputation following endovascular and surgical revascularization was reduced in diabetic and diabetes-free patients alike. The cumulative incidence of death was also reduced (Figure 5).

• Download figure
• Download PowerPoint

The cumulative incidences of amputation and death, by indication for revascularization, are summarized in Figures 6 and 7, and shown in detail in the Appendix in the online-only Data Supplement and Tables X and XI in the online-only Data Supplement. In comparison with 2006 to 2007, the 1-year cumulative incidence of amputation following endovascular revascularization was lower in 2014 to 2015 in all indication categories (Figure 6). The evidence for a decreasing trend was most notable in patients undergoing surgical revascularization for SLI2 or SLI3. In the first group, the cumulative incidence of amputation reduced from 19.5% to 69.5%, from 25.3% to 16.1%. The 1-year cumulative incidence of amputation after endovascular revascularization showed a similar pattern, and the evidence for a decreasing trend in risk was again the clearest in the SLI2 and SLI3 groups (Figure 6 and Table X in the online-only Data Supplement).

• Download figure
• Download PowerPoint

Cumulative incidence of death within a year following both endovascular and surgical revascularization also decreased between the study baseline in 2006 to 2007 and 2014 to 2015, although the pattern was less clear than for amputation (Figure 7). Among patients undergoing endovascular revascularization for IC or SLI1, the risk of death reduced by 2% to 3%, from ≈6% to 3% in the IC group and from 7% to 4% in the SLI1 group. Larger reductions were observed in the SLI2 and SLI3 groups, where the cumulative incidence of death decreased from 20.5% to 14.6% and from 25.5% to 15.9%, respectively (Figure 7 and Table XI in the online-only Data Supplement). After surgical revascularization, the 1-year cumulative incidence of death decreased from ≈10% to 5% in the IC and SLI groups. Again, the reduction was more notable in the SLI3 group (Figure 7 and Table XI in the online-only Data Supplement).

• Download figure
• Download PowerPoint

The proportions of patients undergoing endovascular and surgical revascularization, by year and indication for the intervention, are shown in Figure 8. Overall, the proportion of endovascular revascularizations increased and the proportion of surgical procedures decreased slightly over the study period (P<0.0001, Table and Figure 8). The increasing trend was the clearest among patients undergoing revascularization for SLI3: in this group, the proportion of surgical revascularizations decreased from 33% in 2006 to 2007 to 19% in 2014 to 2015 (P<0.0001). A similar trend was also observed in those treated for IC or SLI2. Among patients in the SLI1 group, the split of the procedures (endovascular and surgical) varied across the study period, but there was little evidence for a trend (P=0.06). Endovascular revascularization also became more common and surgical procedures less common among patients with a record of diabetes mellitus, whereas among those with no diabetes mellitus, the proportions of the 2 types of procedures remained static (Figure III in the online-only Data Supplement).

• Download figure
• Download PowerPoint

Discussion

Summary of Main Findings

Our findings suggest that during our study period, 2006 to 2015, the 1-year risk of major amputation in England decreased from 5.7% to 3.9% following endovascular procedures and from 11.2% to 6.6% following surgical lower limb revascularization. The 1-year risk of death also decreased from 9.5% to 6.0% following endovascular procedures and from 11.1% to 6.4% following surgical procedures. The reduction in the risk of amputation was the largest among patients with the most severe underlying disease (SLI3, severe limb ischemia with a record of gangrene).

We found some evidence that the relative frequency of the 2 types of revascularization procedures changed between 2006 to 2007 and 2014 to 2015. Overall, endovascular revascularization became more common during the study period. The evidence for this trend was clearest among patients who were treated for SLI3 (severe limb ischemia with gangrene). These observations reflect an increase in the overall number of endovascular procedures being performed during the study period, with the number of surgical procedures staying static.

Our investigation focused on 1-year risks of amputation and death following infrainguinal revascularization, and the reduction of 1-year risk of major amputation following revascularization for infrainguinal PAD is in agreement with previous research findings. The BASIL trial (Bypass versus Angioplasty in Severe Ischaemia of the Leg) and a previous study based on HES data, both with outcomes measured in the mid-2000s, reported that ≈12% of patients undergoing (mainly) femoropopliteal bypass underwent major amputation within a year of this operation.^1,6 In our data set, 6.3% of patients undergoing endovascular and 11.9% of patients undergoing surgical revascularization in 2006 to 2007 had an amputation within a year of the index procedure. The slightly lower proportions of amputation outcomes in our analyses may be attributable to different inclusion and exclusion criteria; for instance, we excluded patients undergoing revascularization because of cancer or trauma (≈2% of revascularizations in our data set), which was not done in all previous studies. Our findings should not, however, be interpreted as indicative of the relative merits of endovascular and surgical revascularization.

Several possible explanations for the observed reductions in the 1-year risk of amputation and death exist. One possibility is that the developments in techniques and technology (such as drug-eluting stents) have led to more favorable patient outcomes. Also, during our study period, in particular, the years from 2010 to 2015, vascular services in the United Kingdom were subject to a process of specialization away from general surgery, and centralization from multiple low-volume centers to a smaller number of high-volume specialist centers.¹⁷ It is possible that this reorganization of vascular surgical care has had an impact on the changes and improvements identified in the current study. In response to increasing evidence pointing to a positive relationship between hospital and surgeon volumes and the outcome of arterial surgery,¹⁸ in 2009 the Vascular Society of Great Britain and Ireland published recommendations on reconfiguring the vascular services in the United Kingdom around a hub-and-spoke model. In this model of care, all arterial surgery (including lower limb bypass and major amputations) is centralized into high-volume hub hospitals, with lower-volume spoke hospitals providing local assessment, diagnosis, and less complex interventions. A hub is a hospital that provides a vascular on-call rota of ≥1:6, and serves a population base of at least 800 000.^17,19

Other possible explanations for the decreasing risks of amputation and death among patients undergoing lower limb revascularization for PAD relate to changes in cardiovascular disease (CVD) epidemiology and risk factors in the years leading up to and during the study period, 2006 to 2015. For instance, the prevalence of smoking, an important risk factor for PAD, began to decrease in men and women in all age groups in the United Kingdom in the 1970s, and this trend continued at least to the mid-2010s.^20,21 Coinciding with the reduction in smoking, prescription registers show that statin use increased in the United Kingdom in 1995 to 2013, in keeping with the National Institute of Health and Care Excellence guidelines advising general practitioners (United Kingdom family physicians) to ensure that statins are prescribed to all patients whose 10-year risk of CVD exceeds 20% (>10% since July 2014).^22,23 It is possible that the population-level reductions in smoking exposure and cholesterol levels have led to the severity of PAD decreasing in patient cohorts presenting for revascularization, which in turn could have improved outcomes of these procedures. However, our findings for 2006 to 2015 suggest that the overall proportion of patients having revascularization for milder PAD (marked by IC) decreased, and the proportions of those treated for PAD with tissue loss increased. It seems thus unlikely that the observed improvements in outcomes reflect the decreasing severity of the underlying disease among patients undergoing revascularization in the past few decades.

Coinciding with the increase in statin use and decrease in smoking, the overall burden of CVD in the United Kingdom has declined since the 1970s.²⁴ Mortality and case-fatality from major CVD outcomes (coronary heart disease and stroke) have declined considerably in all United Kingdom countries.^24,25 The prevalence of CVD has, however, remained relatively static and slightly increased in men and women aged ≥65 years between 2004 and 2015, which Bhatnagar and colleagues²⁵ hypothesize could be a result of the reduced mortality and case-fatality rates. Taken together, these trends suggest that the burden of CVD may be shifting from hard and fatal outcomes (eg, myocardial infarction or stroke) to milder forms of the disease, such as PAD. If this is the case, patients who a few decades ago would have died of a major coronary event or stroke, now have their CHD managed and are accessing health services for PAD. In this scenario, the findings of the present investigation would be even more encouraging, if indeed the outcomes were improving despite older and more severely ill patients with more (cardiovascular and other) comorbidities undergoing lower limb revascularization. Our findings lend some support to this hypothesis, for, although patient age remained static throughout the study period, the number of comorbidities and the severity of the underlying PAD appeared to increase. However, these interpretations warrant caution because improvements in the completeness and accuracy of diagnostic coding in HES may have influenced our findings.

Finally, the observation that the risks of amputation, and death reduced steadily in almost all patient groups, could reflect an overall improvement in care, or it could relate to changes in clinical coding. A trend during our study period toward more complete and specific coding of secondary diagnoses (used to identify indications and comorbidities) could have led to overestimation of the risks of amputation and death during the earlier years in patients with the least severe underlying disease. However, this seems to be an unlikely explanation for the observed results, because the falling risks were observed for the total patient cohort and were the largest among the most severely ill patients. Comparison with studies in other countries and settings would help to assess whether our observations can be attributed to improvements in care, changes in CVD prevalence, incidence and risk factors, quality of the data, or some combination of all of these.

Strengths and Limitations

An important strength of our analyses is that we used a large set of routinely collected patient-level data that capture information on all revascularization procedures conducted in NHS hospitals in England. It is thus unlikely that sample selection or loss to follow-up have significantly biased our findings. Furthermore, a data set of just under 104 000 patients gave our analyses sufficient power to produce precise estimates of the risks of major lower limb amputation and death following endovascular and surgical revascularization. A further strength of our analyses is that we used competing risk regression models to examine the risks of amputation and death separately, which is important for unpicking the associations of revascularizations with these outcomes in a population of older patients with multiple comorbidities. The amount of missing data in our analyses was negligible (<1%).

A limitation in our study is that the diagnostic validity of HES for identifying indications for interventions and comorbid conditions is not ideal. However, a recent systematic review suggests that the accuracy of diagnostic coding in this data set has improved since the mid- to late 2000s.^26,27 Indeed, the reduction in the number of patients with no comorbidities (from 17% in 2006–2007 to 12% in 2014–2015, Table) could point to a larger number of comorbidities being accurately recorded in HES. It is possible that this contributed to some of the changes in outcomes and patient characteristics, such as the increasing prevalence of diabetes mellitus, observed over the study period. The coding of major interventions in HES is generally very accurate. Although it is possible that some revascularization procedures have been incompletely recorded or omitted from HES, which might have introduced bias to our estimates, we would expect the size of such bias to be small in comparison with the observed changes in risks.

HES is a rich source of patient-level data on hospital admissions and procedures, but it does not contain data on patient physiology or anatomy, and we were thus unable the gauge the potential confounding effects of patient fitness or vascular anatomy on the selection of patients for different types of revascularization procedures. Finally, we conducted a large number of comparisons, and it is therefore possible that some of the observed associations were chance findings.

Implications and Future Directions

Based on a large set of routinely collected observational data, the findings presented here should be interpreted as descriptive of the care and patient outcomes in England nationwide; further research, however, would be needed to produce risk models that would predict an individual patient’s risk of amputation or death associated with undergoing lower limb revascularization. Large, well-conducted prospective studies, based on disease or procedure registers with detailed information on physiological and anatomic covariates and linked to national hospitalization and death records, would provide the best setting for future research into patient outcomes following lower limb revascularization. They would provide generalizable information, with appropriate statistical power, to investigate the roles of the underlying disease and patient clinical characteristics on the patient care pathway and outcomes.

Over a 10-year period there has been a shift in the pattern of revascularization procedures for infrainguinal PAD in England, with an increasing number of endovascular procedures being performed, especially among patients with the most severe forms of lower limb PAD. The overall survival increased, and the rate of major lower limb amputation decreased over the same period for both endovascular and surgical procedures, despite higher morbidity and larger proportions of patients treated for the severe end of the spectrum of PAD. These trends suggest overall improvements in the outcomes for patients with severe PAD during a period of centralization and specialization of vascular surgical services in the United Kingdom.

Footnotes