Article Text
PDF
Abstract
Background Low socioeconomic position (SEP) has been demonstrated to negatively influence outcome in several cardiovascular patient groups. The aim of this study was to analyse time trends of incidence of intact abdominal aortic aneurysm (iAAA) and ruptured AAA (rAAA), respectively, and to investigate whether SEP had any influence on the probability to present with rupture and, finally, to determine the impact of SEP on outcome.
Methods Nationwide population-based study including all individuals with iAAA or rAAA in Sweden during 2001–2015.
Results The number of individuals with an AAA was 41 222; the majority were identified as iAAA 33 254 (80.7%) and 7968 (19.3%) as rAAA. Time trends showed decreasing incidence of rAAA but increase in iAAA during the study period. Individuals with low income or low educational level were more likely to present with a rAAA rather than iAAA: OR 2.16 (95 % CI 1.98 to 2.36, p<0.001) and OR 1.33 (95 % CI 1.21 to 1.46, p<0.001), respectively. Low income was also associated with increased 90-day mortality and 1-year mortality after treatment for rAAA, OR 1.42 (95% CI 1.07 to 1.89, p=0.014) and OR 1.39 (95% CI 1.13 to 1.97, p=0.005).
Conclusion This large nationwide study showed a decreasing incidence of rAAA. Individuals with low SEP were found to have an augmented risk of presenting with rAAA rather than iAAA and, in addition, to fare worse after repair. Consequently, SEP should be regarded as a relevant risk factor that should be included in considerations for improved care flow of patients with AAA.
- socio-economic
- social inequalities
- epidemiology of cardiovascular disease
- vascular disease
Statistics from Altmetric.com
Introduction
An abdominal aortic aneurysm (AAA) is an asymptomatic disease affecting approximately 1%–2% of all 65-year-old men.1–3 The widened abdominal aorta may expand and possibly lead to rupture, a lethal condition demanding imminent surgical intervention. A significant transformation in the management and care of patients with AAA has occurred during the last 15–20 years. A shift in surgical methods for repairing AAA, from open surgical repair to the expanding use of endovascular aortic repair (EVAR) in both intact AAA (iAAA) as well as in ruptured AAA (rAAA), has broadened the patient eligibility for treatment and possibly improved mortality rates after treatment. In addition, nationwide screening programmes have been implemented and proven to be both cost-effective and durable.3 4 Moreover, changes in sociodemographic and social trends, such as a decrease in smoking,5 has led to a decrease in cardiovascular disease (CVD) and in the iAAA and rAAA incidence.6
Many of the conventional risk factors for developing AAA are already well known and studied (including old age, smoking, male sex and comorbidity),7 8 but the association between socioeconomic position (SEP) and outcome of AAA has only been sparsely investigated.
A clear association between SEP, as defined by disposable household income and level of education, and overall health is well established, but the mechanism through which SEP mediates health is only partly known.9 10 A large study from the USA showed an association between lack of medical insurance and the risk of presenting with a rAAA11 rather than iAAA. SEP has also been demonstrated to affect the choice of treatment modality; a larger proportion of patients with high SEP are treated with EVAR.12 There are, however, still large gaps in the current field of knowledge on the possible influence of SEP on AAA disease.
The aim of this study was to describe long-term trends in incidence and management of rAAA and to compare the SEP distribution in patients with iAAA and rAAA. A particular aim was to elucidate the influence of SEP on type of index diagnosis at presentation (iAAA or rAAA) and on the outcome after rAAA treatment.
Methods
Study population and data sources
Sweden has a population of 10.1 million people (Statistics Sweden, SCB, November 2017), inhabiting an area of 447 435 km2. Elective care for vascular patients is provided by 30 hospitals including seven university hospitals, which are tertiary referral centres. All hospitals have a compulsory reimbursement registry policy. Thereby, the coverage of vascular interventions in national hospital registries is excellent.
All patients diagnosed with an iAAA and/or rAAA between 1995 and 2015 in Sweden were identified through the Swedish National Patient Registry (NPR) including both outpatient clinics and in-hospital care registries, and the Cause of Death Registry (CDR). The registries are maintained by the National Board of Health and Welfare (NBHW, Socialstyrelsen).
The first registered index diagnosis of either iAAA or rAAA was identified for each individual (International Classification of Disease; ICD-9 and ICD-10, respectively) (online supplementary appendix). To ensure that this event indeed represented the first presentation of disease (index event), patients who were diagnosed with an AAA during the period 1995–1999 were excluded. This allowed for a minimum of 5 years (1995–1999) of diagnosis-free interval.
Supplemental material
Patient characteristics including age, sex and comorbidities (registered in the NPR as pre-existing conditions, including ischaemic heart disease, chronic obstructive pulmonary disease (COPD), diabetes mellitus, hypertension, hyperlipidaemia, stroke, renal failure and malignancy (ICD codes appear in online supplementary appendix) were extracted from the NPR. Data on treatment were extracted in a similar manner. The Swedish NPR has a positive predictive value of 85%–96% and is considered to give an accurate depiction of patient-coupled diagnosis and potential treatments at each care event.13 The CDR was used to identify cases with index event not recorded in the NPR as well as the date of death (from all causes) of all included individuals. To allow for all included patients with rAAA to have data on 1-year mortality rates, information on date of death was extracted until 31 December 2016. For the purpose of describing time trends, the study period was divided into five periods, 2001–2003, 2004–2006, 2007–2009, 2010–2012 and 2013–2015. Information on SEP (disposable household income and level of education) was retrieved from Statistics Sweden: statistics and the longitudinal integration database for health insurance and labour market studies (Swedish acronym: LISA). Coupling between different registries was made possible as all individuals in Sweden have their person-specific identity number unique to only them. The merging of extracted data from central registries is performed by the NBHW. Once merged, the anonymised data file, without person-specific identity numbers, is sent to the research unit for analysis. The NBHW follows EU regulations, and the standardised process was approved by the local ethics committee.
Socioeconomic position
Disposable household income, adjusted for household composition (online supplementary appendix) and divided into quintiles, was used as one of two indicators of SEP, with the lowest quintile being referred to as Q1 and the highest as Q5. Each individual was assigned a quintile according to the income distribution in the whole included population and remained categorised within the quintile throughout all analyses. The adjustment for household composition, issued by the Swedish NBHW was performed to accurately assess the disposable household income. This adjustment takes into account the number of individuals sharing a household as well as the number of minors in the household. The disposable household income during the year before the index event was used in the analyses in order to assess a full year of disposable income. Therefore, only individuals with an index event during 2001–2015 were included in the analysis. From records of number of years in school, the levels of education were defined as low (<10 years), middle (10–12 years) and high (>12 years). In all analysis, Q5 was used as reference for disposable household income and the highest educational level was used as reference for level of education.
Outcome
Outcomes were defined as follows: iAAA or rAAA at presentation and 90-day and 1-year mortality after treatment for rAAA.
Statistical methods
Categorical variables were described with frequencies and percentages, and continuous variables with mean and SD. Disposable household income was divided into quintiles. Comparisons of categorical variables were performed using the χ2 test. Continuous variables were analysed using analysis of variance.
Logistic regression was performed to assess the association between SEP and the outcome events. ORs with 95% CIs and correlated p value were calculated, with p value <0.05 considered statistically significant. Highest quintile for disposable household income (Q5) and highest category of level of education (>12 years), respectively, were used as references. For analysis of time trends, a Mantel-Haenszel test for trend was performed using year of event as a continuous variable.
Both crude and adjusted results were presented with SEP as two independent variables, that is, disposable household income and level of education were entered in the model separately. Adjustment was made for age, sex and comorbidity factors. Analysis for missing data >5% was proceeded with multiple imputation by chained equations. Data management and statistical analyses were performed using SAS software V.9.2 (SAS Institute) and SPSS V.25.0 (IBM).
Results
Patients diagnosed with AAA 2001–2015
From 2001 to 2015, a total of 41 222 individuals were diagnosed with either an iAAA (n=33 254; 80.7%) or a rAAA (n=7968; 19.3%) and were included in the analysis. The majority were registered in the NPR, whereas 4597 individuals (11.2%) were only found in the CDR. Patient demographics are presented in table 1. The overall proportion of women was 21.7%. The proportion of women was highest in Q1 and declined by increasing quintile, table 1, p<0.001. The highest proportion of rAAA was found in the poorest quintile (Q1) where 27.1% of the AAA index events were rAAA, whereas only 11.1% of the index events in the richest quintile (Q5) were registered as rAAA, table 1, p<0.001. During the study period, 2761 ruptures occurred among those with a previous index event of iAAA. These cases represent 25.7% of all ruptures during 2001–2015 and were only included in reporting of time trends (table 2). Data on level of education were missing in 14.1% of the included individuals and multiple imputation was performed for this variable. The distribution of comorbidity was slightly skewed with all comorbidity factors (except COPD and stroke) over-represented in the third and fourth quintiles (table 1).
Demographics of included individuals (n=41 222) divided by quintile of disposable household income; Q1 representing the poorest and Q5 the richest quintile
Absolute numbers of AAA cases and mode of presentation per 3-year period
Time trends for iAAA/rAAA
There were 5225 cases of iAAA in the earliest time period (2001–2003) compared with 7304 in the latest (2013–2015) (table 2). The incidence of iAAA per 100 000 inhabitants aged >45 years went from 45.9 in 2001 to 51.4 in 2015 and a peak of 73.5 in 2011 (figure 1A).
Time trends showed a steady decrease in the absolute number of cases of rAAA, from 2505 in the earliest time period to 1679 in the latest (p<0.001) (table 2). This represents a decrease in incidence of rAAA from 22.1 (per 100 000 inhabitants aged ≥45) in 2001 to 11.7 in 2015 (figure 1A). Mean age for index rAAA individuals was 75.7 in 2001 and increased to 78.6 in 2015 (p<0.001). The number of treatments per time period for rAAA declined while treatment rates remained similar (table 2). The use of EVAR for rAAA increased, especially during the last 9 years (2007–2015) (table 2, figure 1B). Postoperative mortality after treatment for rAAA declined during the study period (figure 2).
(A) To the left, annual incidence of intact abdominal aortic aneurysm (iAAA) (blue line) and ruptured AAA (rAAA) (red line) per 100 000 inhabitants aged ≥45. Screening for iAAA was initiated in 2006. Nationwide coverage was achieved in 2015. (B) To the right, the annual number of treated rAAAs per time period (blue staples, left axis) and the % treated with endovascular aortic repair (red line, right axis).
Mortality rates in % of index events of treated patients with ruptured abdominal aortic aneurysm. Blue line indicates 90-day mortality, red line indicates 1-year mortality.
Association between SEP and index presentation of disease
A higher risk of presenting with a rAAA at time of index diagnosis was found for both low disposable household income and low level of education (table 3). After adjusting for age, sex and comorbid conditions, the association remained; OR 2.16 (95% CI 1.98 to 2.36, p<0.001) for Q1 and OR 1.33 (95% CI 1.21 to 1.46, p<0.001) for low level of education. Crude and adjusted results for the association between SEP and presentation of disease at index event are presented in table 3.
Associations between socioeconomic position indicators and presenting with ruptured, as opposed to intact, abdominal aortic aneurysm as index event
Association between SEP and mortality of rAAA
Overall 1-year mortality in patients with rAAA was 76.5% (6096 of 7968). Overall treatment rate for all rAAA occurring 2001–2015 was 36.1% (n=3869), but lower for women (26.7%) compared with men (39.1%), p<0.001. A total of 6860 individuals with rAAA were not treated at time of rupture (including those with a previous index event of iAAA). In the 3054 patients treated for an index event of rAAA, 90-day mortality was 36.1% and 1-year mortality was 38.7%. Low disposable household income (Q1 and Q2), adjusted for sex, age and comorbidities, was associated with increased 1-year mortality after rAAA repair; OR 1.49 (95% CI 1.13 to 1.97, p=0005). Low educational level did not influence survival after rAAA repair (table 4).
Association between socioeconomic position indicators and 90-day and 1-year mortality after treatment for ruptured abdominal aortic aneurysm
Discussion
In this large nationwide population-based study of over 41 000 patients with AAA, a low household income as well as a low educational level were shown to be strongly associated with an increased risk of presenting with a rAAA rather than an iAAA at first-time diagnosis and income was associated with a worse outcome when treated for a rAAA, regardless of age, sex and comorbidities. The basic common educational standards in the population could contribute to the lower influence of education on outcome as compared with the influence of income.
This study is one of the largest population-based studies analysing the mode of presentation of AAA disease and, to our knowledge, the first report that has investigated the association between SEP on an individual level and outcome of rAAA. The included individuals with an iAAA and rAAA represented a typical AAA cohort in regards to the distribution of age, sex and rupture rate.14 15 Also, overall and postoperative mortality rates were similar to those reported in other large population-based studies and randomised clinical trials.16–19 The decrease in rAAA incidence found in our study is in accordance with other national and international results6 20 21 and may mainly be attributed to an increase in iAAA repair and decrease in overall AAA incidence.
The time trends in our study showed that the incidence of iAAA and rAAA, respectively, had an almost inverse relation. It is plausible that the more cases of iAAA that are detected, the less ruptures occur. This would be due to possible increase in iAAA repair and also in improved medical management of these patients including smoking cessation and treatment of hypertension. The increase in EVAR for rAAA repair could possibly explain the unchanged frequency of treatment as more elderly and comorbid patients are eligible for treatment with this method.22 This is supported by the higher mean age over time in the rAAA population found in this investigation and by others.5 22 The observed improvement in postoperative mortality after rAAA repair could be an effect of selection bias, but can also be explained by an improved care flow in the management of the patient with rAAA (ie, permissive hypotension, EVAR-first strategies, transportation protocols).
The observed distribution of comorbid conditions over the quintiles of disposable household income was somewhat expected as high-income individuals are screened and diagnosed with hypertension and hyperlipidaemia at a higher degree, whereas low-income individuals are less prone to seek ambulatory23 and preventative care.24 25
The clear association between low SEP and rAAA as index diagnosis remained strong even after adjustment for sex, age and comorbidity, which implies that the organisations in the healthcare system now need to consider SEP if an increased equality in the distribution of healthcare is desired. Of special significance, only the richest quintile was protective in the analysis with all other quintiles associated with a higher risk of presenting with a rAAA.
We and others have found that low SEP negatively affects the adherence to the nationwide screening programme for AAA disease offered to all 65-year-old men in Sweden.26 27 The association between low SEP and rAAA as index diagnosis has previously been described in a limited number of studies,28 but unlike our study, these only included patients who underwent repair for their rAAA,28 which could lead to an underestimation of the true association between low SEP and presentation of disease.
Studies from North America have suggested a correlation between high SEP and increased rates of treatment by way of EVAR.12 29 An American study of 35 593 patients with AAA also demonstrated an association between low SEP and rAAA as index diagnosis, but could only use insurance status as a marker for SEP.11 The difference in reimbursement systems makes comparisons difficult.
Interestingly, the level of education was only associated with the outcome of type of index diagnosis. For mortality after treatment, no association was seen with the individual’s level of education. This is consistent with outcome rates after other surgical procedures30–32 and indicates a difference in the way socioeconomic factors affect health, with level of education possibly being more influential in comprehending medical information but not as influential on the postoperative periods and overall morbidity. SEP is an intricate composition of different known and unknown factors involved in quantifying an individual’s ability to gain access to healthcare and withstand a course of illness. No single socioeconomic factor alone can fully explain the way in which SEP mediates health, and different index classifications struggle in demonstrating a true representative level of SEP.33 However, by studying these factors separately on an individual basis in a country where all healthcare is tax-funded and regulated by law to be equally distributed to its citizens, one might come a step closer to distinguishing the true effects of SEP on health. An increased awareness of the multifactorial nature through which SEP mediates an individual’s ability to adequately access healthcare, comply with prescribed medicine and live in a health-promoting environment must be developed. These factors are often attributed to income, while understanding disease pathology and adhering to health directives often correlates with level of education.34 35
Strength and limitations
This nationwide study with patient-specific analysis on >41 000 individuals with AAA during a 15-year period is unique, with a complete national coverage. All vascular patients nationwide were included as there are no private caregivers providing AAA repair. The retrospective aspect of our data collection comes with its inherent difficulties and limitations. Although the Swedish NPR has a positive predictive value for CVD of 96% and is considered to give an accurate depiction of patient-coupled diagnosis and potential treatments at each care event,13 there is a risk of misclassification. However, misclassification of disease and treatment is likely to be evenly distributed. The ability to measure SEP on an individual level parallel to a public, tax-funded healthcare registered on individuals, provided to all citizens in Sweden, gives us a unique possibility to elucidate the true effect of SEP on the studied outcomes. Unfortunately, due to the inherent limitation of register-based studies, data on smoking were not possible to attain. The strong association between smoking and AAA has been extensively reported, specifically the negative influence of smoking on risk of development, increased growth and rupture risk.1 7 8 14
As for all registry-based materials on rAAA, there is an unknown number of individuals with rAAA who die out-of-hospital and remain undiagnosed.36–38 Autopsy rates in Sweden have declined over the years and are unlikely to be associated with SEP.
Conclusion
Socioeconomic deprivation has been shown to be detrimental for several cardiovascular patient groups. This large nationwide study confirms a decreasing incidence of rAAA, but also that individuals with low SEP had an augmented risk of presenting with rAAA and suffer from worse outcome. Consequently, SEP should be regarded as a relevant risk factor that should be included in considerations for improved care flow of patients with AAA, such as individualised follow-up, intensified, but easier accessible information to at-risk groups and possibly dedicated screening in specific SEP risk groups.
What is already known on this subject
Socioeconomic position has been shown to influence severity of disease and mortality in the fields of cardiovascular disease and cancer. Incidence of ruptured abdominal aortic aneurysm has decreased in the last few years but remains a lethal condition demanding further understanding of its epidemiology.
What this study adds
This is the first nationwide paper reporting on an association between individual socioeconomic position (SEP) and abdominal aortic aneurysm disease with specific analysis on patients with rupture, and confirms an association on mode of presentation and outcome by SEP, even after adjustment for age, sex and comorbidity. Also, with over 41 000 individuals included in the analysis, it gives a current description of abdominal aortic aneurysm epidemiology and time trends in incidence of rupture, treatment frequencies and changes over time in mortality after treatment.
Acknowledgments
We thank Max Vikström for his support in data management and statistical support.
References
Footnotes
Contributors Conception and design: SZ, RH, KL. Analysis and interpretation: SZ, RH, KL. Data collection: SZ, RH, KL. Statistical analysis: SZ. Writing the manuscript: SZ, RH. Critical revision: SZ, RH, KL, JR, JS. Approval of manuscript and agreement of content: SZ, RH, KL, JR, JS.
Funding RH, SZ and JR had financial support from Swedish Heart–Lung Foundation and through the regional agreement on medical training and clinical research (ALF) between the Stockholm County Council and Karolinska Institutet for the submitted work. JR had financial support from the Stockholm City Council.
Competing interests None declared.
Patient consent Not required.
Ethics approval This study was approved by the Regional Ethics Review Board in Stockholm (registration no. 2015/2108-31/5) and complies with the Declaration of Helsinki.
Provenance and peer review Not commissioned; externally peer reviewed.