Article Text
Abstract
Objectives To investigate the postoperative adverse outcomes among surgical patients with preoperative systemic lupus erythematosus (SLE) in a nationwide population-based study.
Methods We used Taiwan's National Health Insurance Research Database to identify 4321 surgical inpatients with SLE and 17 284 sex- and age-matched controls receiving major surgery. Sociodemographic characteristics, preoperative comorbidities, postoperative 30-day in-hospital major complications and mortality were analysed among surgical patients with and without SLE.
Results Surgical patients with SLE had a higher prevalence of preoperative coexisting medical conditions and postoperative major complications. The OR of 30-day postoperative mortality for surgical patients with SLE was 1.71 (95% CI 1.09 to 2.67) after adjustment. Surgical patients who had received more recent (within 6 months) preoperative SLE-related inpatient care had higher risks of 30-day postoperative acute renal failure (OR=7.23, 95% CI 4.52 to 11.6), pneumonia (OR=2.60, 95% CI 1.82 to 3.72), pulmonary embolism (OR=4.86, 95% CI 1.20 to 19.7), septicaemia (OR=3.43, 95% CI 2.48 to 4.74), stroke (OR=2.01, 95% CI 1.38 to 2.92), overall complications (OR=2.30, 95% CI 1.89 to 2.80) and 30-day postoperative mortality (OR=2.39, 95% CI 1.28 to 4.45) than surgical patients without SLE. SLE-related preoperative steroid injections showed a dose-dependent relationship with postoperative complications and mortality.
Conclusions SLE significantly increased the risks of surgical patients for overall major complications and mortality after major surgery. Our findings demonstrated the need for integrated care and revised protocols for perioperative management to improve outcomes for surgical patients with SLE.
- Systemic Lupus Erythematosus
- Outcomes research
- Health services research
Statistics from Altmetric.com
Introduction
Systemic lupus erythematosus (SLE) is an autoimmune connective tissue disorder with a broad range of clinical presentations that predominantly affects Asian and African women.1 In addition to a higher prevalence of SLE in oriental people,2 the burden of SLE is considerably increased among non-white racial groups.3 In Asia, it has an annual incidence of 0.9–3.1 per 100 000 people and a prevalence rate of 30–50 per 100 000 people.4 Active lupus, severe infections and cardiovascular diseases are the leading causes of SLE death.5 SLE exhibits a bimodal mortality pattern6: early death mainly results from active lupus and severe infections in patients receiving high-dose steroids, and late death from steroid-precipitated atherosclerotic cardiovascular diseases.6–8 Disease status and complications secondary to steroid use are thus major determinants of mortality in patients with SLE. Hospitalisation due to SLE has been used as a surrogate indicator of active SLE in a previous investigation of short-term prognosis,9 and a morbidity measure in patients with SLE.10 The injection of large-dose steroids, which might lead to related postoperative complications,11 is commonly used in severe organ-affected or life-threatening SLE.12 From these perspectives, the frequency of steroid injections during prior hospitalisation might be considered as a clinical predictor of postoperative complications in patients with SLE.
However, no large-scale analysis of the global features of perioperative morbidity and mortality in surgical patients with SLE has been carried out. We attempted to clarify whether SLE independently leads to postoperative major complications after inpatient surgery. We also investigated whether the frequency of steroid injections before hospitalisation for index surgery predicts the occurrence of postoperative adverse outcomes.
Methods
Data sources
Research data were obtained from the reimbursement claim of the universal National Health Insurance programme which was implemented in Taiwan in March 1995, covering more than 99% of 22.6 million Taiwan residents. Taiwan's National Health Research Institutes established a National Health Insurance Research Database recording all beneficiaries’ medical services, including inpatient and outpatient demographics, primary and secondary diagnoses of diseases, procedures, prescriptions and medical expenditure. The validity of the Taiwan National Health Insurance Research Database has been evaluated and research articles have been accepted world-wide for public access.13–15
Ethical approval
Insurance reimbursement claims used in this study were from Taiwan's National Health Insurance Research Database, which is available for public access. To protect personal privacy, the electronic database was decoded with patient identifications scrambled to enable further public access for research. According to National Health Research Institute regulations, informed consent is not required as patient identification has been decoded and scrambled. However, this study was still evaluated and approved by Taiwan's National Health Research Institutes.
Study population
We examined medical claims and identified 4321 patients aged ≥20 years with a preoperative diagnosis of SLE from 2 010 412 people of all ages (including 1 336 384 adults aged ≥20 years) who underwent major inpatient surgery, defined as those requiring general, epidural or spinal anaesthesia and hospitalisation for more than 1 day between 2004 and 2007 in Taiwan. At least two claims for patients with the principal diagnosis of SLE within the 24-month preoperative period were requested. For 4321 surgical patients with SLE, we randomly selected 17 284 subjects from surgical populations free from SLE, matched by sex and age with frequency matching, for analysis.
According to the regulations of Taiwan's National Health Insurance programme, insured beneficiaries who have catastrophic diseases can apply for a catastrophic diseases certificate. If this is issued medical co-payment is completely waived for these patients. The Bureau of National Health Insurance strictly reviewed the application and examined the applicators’ medical records, laboratory and image studies. SLE is defined as a catastrophic disease in Taiwan. People who had had two medical visits to a doctor with a primary diagnosis of SLE were considered eligible SLE cases qualifying for a catastrophic diseases certificate. The criteria for SLE were also used in a previous study.16
Outcome measures
Background parameters, such as whether medical services were provided by a teaching hospital, were analysed. Low income was defined as occurring in patients qualifying for totally waived medical co-payment, as certified by the Bureau of National Health Insurance. Some patients with catastrophic disease might not have had a low income. Urbanisation was estimated from the population density (people/km2), defined by dividing the 359 townships and city districts’ populations by the area (km2) of each administrative unit and categorised into four quartiles: low, moderate, high and very high urbanisation. In accordance with the International Classification of Diseases, 9th revision, clinical modification (ICD-9-CM), we defined preoperative SLE (ICD-9-CM 710.0) and major coexisting medical illnesses including hypertension (ICD-9-CM 401-405), chronic obstructive pulmonary disease (ICD-9-CM 490-496), diabetes mellitus (ICD-9-CM 250), stroke (ICD-9-CM 430-438), acute myocardial infarction (ICD-9-CM 410), congestive heart failure (ICD-9-CM 428), peripheral vascular disease (ICD-9-CM 443) and acute renal failure (ICD-9-CM 584) as being recorded from medical claims for the 24-month preoperative period.13 Renal dialysis was also considered as preoperative functional status. Eight major postoperative complications were monitored: acute myocardial infarction, acute renal failure, deep-wound infection (ICD-9-CM 958.3), pneumonia (ICD-9-CM 480-486), postoperative bleeding (ICD-9-CM 998.0, 998.1 and 998.2), pulmonary embolism (ICD-9-CM 415), septicaemia (ICD-9-CM 038 and 998.5) and stroke. These complications and subsequent overall in-hospital mortality within 30 days after the index surgery were the study's primary outcomes.13 ,17–20 Because our study focused on the association between SLE and 30-day in-hospital postoperative mortality, patients who died on the 31st day or later of hospitalisation were counted as alive in this study. Patients who died out of hospital within 30 days were not counted in the mortality outcome. This definition of 30-day in-hospital postoperative mortality was also used in our previous studies.13 ,19 ,20
To investigate the potential impact of clinical severity of SLE on postoperative adverse outcomes, we evaluated the history of hospitalisation in patients with SLE as the primary diagnosis within the 24-month preoperative period and analysed the dose–response relationship of steroid injections with postoperative complications and mortality during the same period.
Data analysis
We used χ2 tests to analyse descriptive parameters for demographic status and coexisting medical conditions in a comparison of postoperative complications and death rates of surgical patients with and without SLE. Continuous variables were analysed with t tests to compare differences between SLE cases and controls.
Odds ratios (ORs) with 95% confidence intervals (CIs) for 30-day postoperative complications and mortality between surgical patients with or without SLE were analysed with multivariate logistic regression by adjusting for sex, age, operation in teaching hospital or not, preoperative coexisting medical conditions and types of surgery and anaesthesia. To verify the impact of disease severity on inpatient adverse outcomes within 30 days, including readmission during the period after the index surgery, multivariate logistic regression was used to control the above covariates and to analyse the history of hospitalisation due to SLE and steroid injections during prior hospitalisation within 24 months preoperatively. Statistical analysis system software V.9.1 (SAS Institute, Cary, North Carolina, USA) was used for data analyses; differences between groups were considered significant if two-sided p values were <0.05.
Results
A higher proportion of surgical patients with SLE received operations in teaching hospitals compared with surgical patients without SLE (94.1% vs 83.1%, p<0.0001) (table 1). Compared with patients without SLE, surgical patients with SLE were more likely to be low-income status (3.3% vs 2.6%, p=0.0082) and have a higher prevalence of preoperative coexisting medical conditions such as hypertension (19.6% vs 18.3%, p=0.0479), chronic obstructive pulmonary disease (11.5% vs 8.7%, p<0.0001), rheumatoid arthritis (9.6% vs 1.2%, p<0.0001), renal dialysis (5.9% vs 1.1%, p<0.0001), osteoporosis (4.1% vs 2.6%, p<0.0001), stroke (2.3% vs 1.8%, p=0.0223), congestive heart failure (1.9% vs 1.3%, p=0.0024), peripheral vascular disease (1.6% vs 0.4%, p<0.0001), hypothyroidism (1.3% vs 0.5%, p<0.0001), myocardial infarction (0.8% vs 0.4%, p=0.0002) and acute renal failure (0.7% vs 0.2%, p<0.0001). Significant differences between surgical patients with and without SLE were also found in types of surgery (p<0.0001) and anaesthesia (p<0.0001).
Patients with SLE showed higher rates of 30-day postoperative complications, including acute renal failure (1.4% vs 0.5%, p<0.0001), pneumonia (2.2% vs 1.4%, p<0.0001), pulmonary embolism (0.2% vs 0.1%, p<0.0001), septicaemia (2.9% vs 1.5%, p<0.0001) and overall complications (8.9% vs 6.6%, p<0.0001) (table 2). The 30-day postoperative mortality rates for surgical patients with and without SLE were 0.8% and 0.4%, respectively (p<0.0001).
After adjusting for sex, age, teaching hospital or not, low income, urbanisation, coexisting diseases, types of surgery and anaesthesia, our multivariate logistic regression analyses found that surgical patients with SLE had significantly higher risk of postoperative mortality (OR=1.71, 95% CI 1.09 to 2.67) and major complications including acute renal failure (OR=2.90, 95% CI 2.01 to 4.17), pneumonia (OR=1.31, 95% CI 1.01 to 1.70), septicaemia (OR=1.75, 95% CI 1.38 to 2.21) and overall complications (OR=1.29, 95% CI 1.13 to 1.47) (table 2).
Further analysis of correlation between postoperative adverse outcomes and clinical severity of SLE was evaluated through patients’ use of medical resources such as preoperative hospital care for SLE (table 3) and steroid injections for SLE treatment (table 4). Compared with the non-SLE group, surgical patients who received more recent inpatient care for SLE (within 6-month preoperatively) were more likely to have increased risk of postoperative acute renal failure (OR=7.23, 95% CI 4.52 to 11.6), pneumonia (OR=2.60, 95% CI 1.82 to 3.72), pulmonary embolism (OR=4.86, 95% CI 1.20 to 19.7), septicaemia (OR=3.43, 95% CI 2.48 to 4.74), stroke (OR=2.01, 95% CI 1.38 to 2.92), overall complications (OR=2.30, 95% CI 1.89 to 2.80) and 30-day postoperative mortality (OR=2.39, 95% CI 1.28 to 4.45) after adjusting for age, sex, teaching hospital, low income, urbanisation, coexisting medical conditions, types of surgery and types of anaesthesia.
After adjusting for age, sex, teaching hospital, low income, urbanisation, coexisting medical conditions, types of surgery and types of anaesthesia in the multivariate logistic regression, dose-dependent relationships were found between steroids injections and postoperative adverse outcomes including acute myocardial infarction (OR=2.90, 95% CI 1.08 to 7.82), acute renal failure (OR=5.15, 95% CI 2.89 to 9.17), pneumonia (OR=3.59, 95% CI 2.44 to 5.30), septicaemia (OR=4.23, 95% CI 2.92 to 6.13), overall complications (OR=2.73, 95% CI 2.14 to 3.50) and 30-day mortality (OR=2.86, 95% CI 1.38 to 5.93) among surgical patients with SLE.
Of patients undergoing renal dialysis (see online supplementary table S1), those with SLE were at higher risk of postoperative mortality than those without SLE (OR=1.78, 95% CI 1.12 to 2.83) after adjustment for age, sex, teaching hospital, low income, urbanisation, coexisting disease, types of surgery and types of anaesthesia. Online supplementary table S2 shows the association between SLE and postoperative adverse events stratified by types of surgery and types of anaesthesia in the multiple logistic regressions analyses with the adjustment for age, sex, teaching hospital, low income, urbanisation and coexisting disease.
Discussion
This retrospective, nationwide, population-based cohort study showed that by using the history of hospitalisation due to SLE and steroid injections during prior hospitalisation as indicators of SLE severity, risks of surgery/anaesthesia can be classified accordingly. Most of our results pertained only to the subgroups with evidence of these histories and therefore not to the SLE group as a whole. The more recent history of hospitalisation due to SLE and the more frequent steroid injections before hospitalisation for index surgery, the higher risks of postoperative 30-day in-hospital overall complications and mortality. Compared with other complications, septicaemia (adjusted OR=24.3) and acute renal failure (adjusted OR=16.4) were associated with the highest death rate (not shown in the tables) and their risks correlated most consistently with SLE severity, and with 30-day mortality. Although 8.9% of patients with SLE receiving surgery developed postoperative complications, the 30-day postoperative mortality among them was found to be low, but significant. Thus about 1 in 125 surgical admissions had a fatal outcome. The SLE mortality risk estimate for surgical hospitalisation is lower than that for all-cause hospitalisation.21 Our result showed a high proportion of men among the patients with SLE, an intriguing fact consistent with previous cohort studies in Taiwan.22 ,23 Similar to the concurrence of rheumatoid arthritis and SLE reported in an Australian case series,24 an increased prevalence of rheumatoid arthritis was found among our lupus patients.
It might be argued that postoperative complications are confounded by the direct effect of steroids. Although steroids themselves may have their own side effects, a preoperative single-dose steroid injection during surgical hospitalisation does not obviously cause postoperative complications in cases of inpatient major surgery.25 Furthermore, steroid injections before, but not during, hospitalisation for index surgery in our study mainly reflected SLE severity and helped to exclude the impact of steroid-induced postoperative complications.
A risk of septicaemia, but not deep wound infection, increased in patients with SLE postoperatively, indicating intrinsic immune dysfunction rather than an extrinsic infectious source. Although there has been a dramatic decrease in SLE death due to infection over recent decades,26 sepsis still predicts short-term mortality.9 Patients receiving higher doses of steroid for disease control are at higher risk of developing postoperative septicaemia.27 There is no consensus about antibiotic prophylaxis in patients with SLE undergoing invasive procedures.28 Our results showed a graded risk of postoperative infectious complications and may provide valuable information for making clinical decisions and developing perioperative prophylactic guidelines for surgical patients with SLE.
Pneumonia is the most common type of infection resulting in SLE mortality in oriental populations.29 ,30 The dosage of steroids before final admission is a surrogate indicator for SLE severity, and also an independent correlate of death from infection, especially for fatal opportunistic pneumonia in patients with SLE.31 Our data showed that the risk of pneumonia increased postoperatively in patients with SLE, especially in those receiving multiple steroid administrations (three or more injections) during prior hospitalisation. Although Kinder et al32 found that medication history was unrelated to the occurrence of pneumonia in patients with SLE, their study was limited by mixed ethnicity with low power to examine non-Caucasian groups and unable to assess the impact of corticosteroid dosage on the risk of pneumonia. In contrast, our result focused on a population-based Asian group and assessed the impact of steroids administered by injection in surgical patients with SLE, which dose-dependently increased the risk of pneumonia postoperatively.
For surgical patients with SLE, the risks of acute renal failure (adjusted OR=7.23 for history of hospitalisation within 6 months preoperatively and adjusted OR=5.15 for three or more steroid injections before hospitalisation of index surgery) were high among major postoperative complications in comparison with the risk for surgical patients without SLE. A multivariate analysis has shown that SLE nephropathy correlates with poorer survival.33 In our cohort, acute renal failure was the second highest cause of death (adjusted OR=16.4) among major complications (not shown in the tables). The reasons for postoperative acute renal failure in patients with SLE are not well defined. Our data showed that its occurrence was strongly dependent on the severity of SLE and, therefore, might parallel lupus flare. In addition, diathesis for intravascular coagulation may provide another explanation for postoperative rapid deterioration of renal function in a patient with SLE.34 Thrombotic tendencies in SLE may precipitate glomerular thrombosis,35 which in some cases characterises acute renal failure in patients with SLE35–37; heparin and aspirin have thus been proposed as important adjuvant treatments to save the kidneys.36–38
Preclinical coronary heart disease39 and carotid plaque40 are increased in patients with SLE. The use of corticosteroids, as well as being a marker for more active lupus, may also be atherogenic.5 SLE itself should be added to the list of conditions that raise cardiovascular risk independently of conventional risk factors39 to avoid underestimation.41 This is particularly the case for acute myocardial infarction in surgical patients with SLE who received multiple steroid administrations (three or more injections) before hospitalisation for index surgery. Although carotid plaque is more prevalent among patients with SLE, higher doses of steroids may decrease its likelihood.40 This atherosclerosis-modifying effect of steroids probably affected our results for stroke, showing that the risk of stroke was only increased in patients with SLE had a small number (one or two) of steroid injections, but this was balanced by multiple steroid administrations (three or more injections) preoperatively.
Pulmonary complications seem to occur in patients hospitalised for exacerbations of SLE.42 Therefore, the occurrence of a pulmonary embolism might parallel lupus flare. Our results also showed that more recent (within 6 months) hospitalisation due to SLE predicts the occurrence of postoperative embolism. The thrombotic risk is inherent in patients with SLE,1 ,43 and the incidence of thrombosis in the first six postoperative weeks is over 100 times that without surgery.44 Therefore, when patients with SLE undergo surgery, an additive or synergistic effect on the development of thrombosis is expected, as shown in our results.
Although it is well known that the presence of antiphospholipid antibodies will aggravate the thrombotic tendency in patients with SLE, our analysis could not provide the relevant antiphospholipid results. The ICD-9-CM codes could neither specify the diagnosis of antiphospholipid syndrome, nor provide information as to whether postoperative thrombotic complications were linked to the presence of antiphospholipid antibodies. Furthermore, the intensity of perioperative anticoagulation for SLE with antiphospholipid syndrome remains to be determined,1 ,45 and recent guidelines on reducing the risk of inpatient thrombosis46 have not included patients with SLE. Current recommendations for perioperative thromboprophylaxis in patients with SLE should be re-examined43 in light of our data, at least for postoperative pulmonary embolism and possibly thrombotic renal failure, as mentioned above.
This study is limited by the use of retrospective reimbursement claims, which mainly comprised primary/secondary diagnoses by ICD-9-CM codes without individual detailed biochemical data. The lack of information on clinical features of SLE (details of major organ involvement, SLE duration/damage and use of specific immunosuppressive agents) and on lifestyles (smoking, physical activity and drinking of alcohol) also limits this study. Information about perioperative thromboprophylaxis and the means of steroid injection—bolus or infusion— could not be specified from our database. There is a possibility that information about comorbidities predating the 24-month preoperative period is missing. Error in coding for preoperative comorbidities and postoperative complications is also possible with such a large population-based database; however, the reliability of the National Health Insurance Research Database is acceptable.13–15 Postoperative complications in patients with SLE were excluded if the same comorbidities has been diagnosed preoperatively; therefore, postoperative complications presenting as recurrent comorbidities were underestimated.
We conclude that patients with SLE have higher risks of adverse outcomes after in-hospital major operations. By evaluating the history of prior hospitalisations and further stratifying the frequency of steroid injections, postoperative risks can be easily assessed and predicted. The results also provide clinical information that will help to improve the quality of perioperative care and thus reduce postoperative adverse events for this specific population.
Acknowledgments
This study is based (in part) on data from the National Health Insurance Research Database provided by the National Health Research Institutes. The interpretation and conclusions contained herein do not necessarily represent those of the National Health Research Institutes.
References
Footnotes
-
Handling editor Tore K Kvien
-
Contributors All the authors have approved the manuscript and contributed to the study design, data analysis, interpretation of data and writing of the paper.
-
Competing interests None.
-
Provenance and peer review Not commissioned; externally peer reviewed.