Do Patients Live Longer After THA and Is the Relative Survival Diagnosis-specific?

Abstract

Background

Hip replacements are successful in restoring mobility, reducing pain, and improving quality of life. However, the association between THA and the potential for increased life expectancy (as expressed by mortality rate) is less clear, and any such association could well be influenced by diagnosis and patient-related, socioeconomic, and surgical factors, which have not been well studied.

Questions/purposes

(1) After controlling for birth year and sex, are Swedish patients who underwent THA likely to survive longer than individuals in the general population? (2) After controlling for relevant patient-related, socioeconomic/demographic factors and surgical factors, does relative survival differ across the various diagnoses for which THAs were performed in Sweden?

Methods

Data from the Swedish Hip Arthroplasty Register, linked to administrative health databases, were used for this study. We identified 131,808 patients who underwent THA between January 1, 1999, and December 31, 2012. Of these, 21,755 had died by the end of followup. Patient- and surgery-specific data in combination with socioeconomic data were available for analysis. We compared patient survival (relative survival) with age- and sex-matched survival data in the entire Swedish population according to Statistics Sweden. We used multivariable modeling proceeded with a Cox proportional hazards model in transformed time.

Results

Patients undergoing elective THA had a slightly improved survival rate compared with the general population for approximately 10 years after surgery. At 1 year after surgery, the survival in patients undergoing THA was 1% better than the expected survival (r = 1.01; 95% confidence interval [CI], 1.01-1.02; p < 0.001); at 5 years, this increased to 3% (r = 1.03; 95% CI, 1.03-1.03; p < 0.001); at 10 years, the difference was 2% (r = 1.02; 95% CI, 1.02-1.03; p < 0.001); and by 12 years, there was no difference between patients undergoing THA and the general population (r = 1.01; 95% CI, 0.99-1.02; p = 0.13). Using the diagnosis of primary osteoarthritis as a reference, hip arthroplasties performed for sequelae of childhood hip diseases had a similar survival rate (hazard ratio [HR], 1.02; 95% CI, 0.88-1.18; p = 0.77). Patients undergoing surgery for osteonecrosis of the femoral head (HR, 1.69; 95% CI, 1.60-1.79; p < 0.001), inflammatory arthritis (HR, 1.49; 95% CI, 1.38-1.61; p < 0.001), and secondary osteoarthritis (HR, 2.46; 95% CI, 2.03-2.99; p < 0.001) all had poorer relative survival. Comorbidities and the Elixhauser comorbidity index had a negative association with relative survival. Level of achieved education (middle level of education: HR, 0.90, 95% CI, 0.87-0.93, p < 0.001; high level: 0.76, 95% CI, 0.73-0.80, p < 0.001) and marital status (single status: HR, 1.33; 95% CI, 1.28-1.38; p < 0.001) were also negatively associated with survival.

Conclusions

Whereas it has been known that in most patients, THA improves quality of life, this study demonstrates that it also is associated with a slightly increased life expectancy that lasts for approximately 10 years after surgery, especially among patients whose diagnosis was primary osteoarthritis. This adds further proof of a health-economic value for this surgical intervention. The reasons for the increase in relative survival are unknown but are probably multifactorial.

Level of Evidence

Level III, therapeutic study.

Introduction

The number of electively performed THAs has been increasing worldwide and THA is considered one of the fastest growing surgical interventions, because of its track record in consistently restoring patients’ mobility, reducing pain, and improving health-related quality of life [8, 11, 20, 24, 43]; it also appears to be cost-effective [18]. Further increases in demand are expected [17, 31, 32] and the burden of hip diseases is likely to remain high [10]. Whereas the operation is “life-changing” in most patients, one should still consider, like with all other major planned surgery, the mortality rate and the influence on life expectancy after the operation. There are strong indications that patients’ survival is improving and generally patients undergoing THA have a tendency to live longer than a matched general population [3, 21, 25]. Register studies have shown that patients undergoing THA have lower mortality than a decade ago despite the increased patient complexity and increase in medical comorbidities [25, 46]. Studies highlighted a slightly elevated early mortality rate [2, 5, 6, 13, 21]. Differences in survival rates have been associated with different indications [21, 33].

However, a number of unanswered questions remain, including how, exactly, survival after THA compares with survival in age- and sex-matched individuals from the general population. This is important because we have not yet found the “eternal hip prosthesis” and so a patient’s lifetime risk for revision surgery will depend, to some degree, on that patient’s expected survival [1, 4, 39]. In addition, a patient’s expected survival may influence a surgeon’s choice of implant, bearing, and fixation technique. To our knowledge, long-term survival has only been described for patients with osteoarthritis and without the added benefit of having access to information on socioeconomic and comorbidity status [25]. Finally, potential differences in survival among patients with different preoperative diagnoses that precipitated the THA (such as primary osteoarthritis, inflammatory arthritis, sequelae of childhood hip diseases, and osteonecrosis of the femoral head) have been explored [21, 29] but not definitively characterized. Having more information on this topic might help in the decision-making process on the basis of the underlying diagnosis for the surgical intervention.

The primary purpose of this study, therefore, is to describe the survival rate for a cohort of Swedish patients who underwent elective hip arthroplasty. The survival of this cohort is compared with the observed survival in the entire Swedish population controlling for sex and age corresponding to the relative survival of this THA population. Specifically, we asked: (1) After controlling for birth year and sex, are Swedish patients who underwent THA likely to survive longer than individuals in the general population? (2) After controlling for relevant patient-related, socioeconomic/demographic factors and surgical factors, does relative survival differ across the various diagnoses for which THAs were performed in Sweden?

Patients and Methods

A retrospective analysis of longitudinally maintained information from the Swedish Hip Arthroplasty Register (SHAR) was conducted. All patients undergoing THA in Sweden are entered in the SHAR database. The SHAR is part of the Swedish Quality Registers [12] and is known to have high completeness (98%) and full coverage (100%) [40]. In patients who had died, the date of death was collected from the Swedish Tax authorities.

Ethical review approval was obtained from the Regional Ethical Review Board in Gothenburg, Sweden (decision 271-14). The data set for this study includes only patients undergoing elective THA (not acute hip fractures nor tumor surgery) whom underwent the operation(s) between January 1, 1999, and December 31, 2012, as registered in the SHAR. The most frequent indication for hip arthroplasty was primary osteoarthritis (91% [120,677 of 131,808]). A total of 131,808 patients underwent elective THA. Of these, 21,755 had died by the end of followup. There were differences for most covariates considered between survivors and patients who died (Table 1). The median followup was 5.62 years (interquartile range [IQR], 6.40 years; maximum, 14 years) for survivors and the median followup was 5.43 years (IQR, 5.16 years) for those who died.

Table 1.

Baseline population demographics of the whole cohort, the surviving and the dead cohorts, and p values

We used a flowchart illustrating patients’ selection for this analysis (Fig. 1). Whereas there is a theoretical risk of leakage out of the register, so far only four patients have requested to withdraw their data. The level of emigration in Scandinavian countries is extremely low and, in any event, emigration would have to be communicated to the tax office and as such would be recorded in the SHAR. The SHAR database has always considered patients who have emigrated (and as such lost their personal ID number) as dead. It is impossible within the study database and within the SHAR database to retrace the exact number of emigrated patients.

Fig. 1.

The flow diagram illustrates patient selection for the relative survival analysis of Swedish patients undergoing primary THA.

This longitudinal, comparative study compared survival (that is, patient survival, not implant survival) between patients who underwent THA in Sweden with aggregated survival data at the national level from the general population as calculated by Statistics Sweden (Human Mortality Database [http://www.mortality.org]). The main purpose of the comparison—and the primary study outcome of interest—is to determine whether patients who undergo THA live longer than age- and sex-matched patients who have not undergone THA. The secondary study purpose was to evaluate diagnosis-specific survival (again, patient survival, not implant survival) among patients who have undergone THA. Those comparisons were made here by using the different diagnoses as coded in the SHAR, comparing them with survival data taken from the Human Mortality Database. To study the influence of covariables, data from the SHAR were combined with data from Statistics Sweden and registers under the umbrella of the National Board of Health and Welfare in Sweden. These three main sources of data have been successfully linked with the help of the unique 10-digit personal ID number [22]. In detail, a description of the data acquisition and the merger can be found in a previous article on the linkage of Swedish health data registers [9]. We were able to identify patient- and surgery-specific data relating to elective THAs from the SHAR including data about the date of operation, underlying diagnosis, and fixation technique. Mortality data are obtained by crossmatching with the Swedish population register governed by the Swedish Tax Agency [30]. Level of education characteristics, where we defined low, middle, and high level of education to respectively describe the first 9 years of education (up to the end of the secondary school level), an extra 2 or 3 years (college level) and higher education (university or similar), and civil status are available from Statistics Sweden. The Elixhauser comorbidity index was calculated using the algorithm by Quan and collaborators [38] based on the comorbidities (International Classification of Diseases, 10^th Revision) recorded in the Patient Register of the National Board of Health and Welfare of Sweden. We decided to use the Elixhauser comorbidity index because this has previously been identified as being the most predictive of mortality [7, 41]. Different variables were studied (Table 1).

Statistical Analysis

Continuous variables were summarized as means and SDs and categorical variables as percentages. Group comparisons were run with Student’s t-test and chi-square test. Survival data were summarized as Kaplan-Meier survival curves and relative survival ratios [35-37, 45]. The relative survival ratio is defined as the observed survival in the patient group divided by the expected survival of a comparable group from the general population. The formula is reproduced subsequently:

Where S_O(t) denotes the observed survival in the studied group and S_P(t) is the population or expected survival. The latter was extracted from publicly available mortality tables maintained by the Human Life-Table Database (http://www.lifetable.de/; although this website is owned and maintained by a German-based international organization, we only used the table that contained Swedish data) and Human Mortality Database (http://www.mortality.org). We accessed life tables between 1999 and 2012 tabulated for birth year and sex. Multivariable modeling proceeded with a Cox proportional hazards model in transformed time [44]. Model assumptions were checked with Brownian bridges [45]. We observed significant deviation from the assumption of proportionality for the Elixhauser comorbidity index. We mitigated the problem with introducing time-dependent coefficients. Graphic examination of the effect of the Elixhauser comorbidity index indicated that there are changes in the effect measures at 5 and 8 years. Thus, we introduced a step function that split the data into three epochs: up to 5 years, between 5 and 8 years, and above 8 years. The regression model then included an interaction term between the Elixhauser comorbidity index and step function for time. The hazard rates for the Elixhauser comorbidity index for the different epochs are sums of the main and interaction terms.

All analyses were performed using R Statistical Software (R Foundation for Statistical Computing, Vienna, Austria). A p value < 0.05 was set to define statistical significance. Confidence intervals have a nominal coverage of 95% and we consider a hazard ratio (HR) significant if the associated 95% confidence interval does not include the reference value of 1.

Results

Patients undergoing elective THA had a slightly improved survival rate compared with the general population for approximately 10 years after surgery (Fig. 2). At 1 year after surgery, the survival in patients undergoing THA was 1% better than the expected survival (r = 1.01; 95% confidence interval [CI], 1.01-1.02; p < 0.001); at 5 years, this increased to 3% (r = 1.03; 95% CI, 1.03-1.03; p < 0.001); at 10 years, the difference was 2% (r = 1.02; 95% CI, 1.02-1.03; p < 0.001); and by 12 years, there was no difference between patients undergoing THA and the general population (r = 1.01; 95% CI, 0.99-1.02; p = 0.13). Patients with primary osteoarthritis had better survival than most other diagnoses when matched for age, sex, education, marital status, and comorbidity throughout the followup period (Fig. 2). Using the diagnosis of primary osteoarthritis as a reference, hip arthroplasties performed for sequelae of childhood hip diseases had a similar survival rate (HR, 1.02; 95% CI, 0.88-1.18; p = 0.77). Patients undergoing surgery for femoral head necrosis (HR, 1.69; 95% CI, 1.60-1.79; p < 0.001), inflammatory arthritis (HR, 1.49; 95% CI, 1.38-1.61; p < 0.001), and other types of secondary osteoarthritis (HR, 2.46; 95% CI, 2.03-2.99; p < 0.001) all had poorer relative survival than did patients with primary osteoarthritis (Fig. 3). Comorbidities and specifically the Elixhauser comorbidity index had a negative association with the relative survival at the different epochs (Table 2). Level of achieved education (middle level of education: HR, 0.90, 95% CI, 0.87-0.93, p < 0.001; high level: 0.76, 95% CI, 0.73-0.80, p < 0.001) and marital status (single status: HR, 1.33; 95% CI, 1.28-1.38; p < 0.001) were also negatively associated with survival. Females had generally better survival than males (HR, 0.97; 95% CI, 0.94-1.00; p = 0.025). Patients operated on at more advanced ages had an overall better relative survival than their peers of the same birth year from the general population (HR, 0.96; 95% CI, 0.96-0.96; p < 0.001). Of the different fixation types, only the uncemented fixation type differed from the cemented type (reference) and uncemented fixation was associated with better survival (HR, 0.78; 95% CI, 0.70-0.87; p < 0.001) (Table 2). Hybrid (cemented stem/uncemented cup) (HR, 0.93; 95% CI, 0.83-1.03; p = 0.15) and use of reverse hybrid hips (uncemented stem/cemented cup) (HR, 0.93; 95% CI, 0.83-1.04; p = 0.18) had no statistically significant association with relative survival (Table 2). The year of operation was positively associated with survival rate with patients operated on in the later years of the study period showing better survival (HR, 0.95; 95% CI, 0.95-0.96; p < 0.001) (Table 2). Thus, patients operated on in the later years had better short-term survival than those operated on in the beginning of the observation period. No conclusions could be drawn regarding longer time survival because of the short observation period for patients operated on during the last years of the observation period.

Fig. 2.

Relative survival of the cohort is presented up to 14 years postoperatively. The horizontal gray line represents the relative survival rate of 1 when the survival of patients undergoing hip arthroplasty equals the survival rate of their sex- and birth year-matched peers from the general population. Deviation from the reference line represents better or worse survival (patient survival).

Fig. 3.

Relative survival is presented for patient groups based on diagnostic indications for THA (chi square = 321.5, df = 4, p < 0.001). The horizontal gray line represents the relative survival rate of 1 when the survival of patients undergoing hip arthroplasty equals the survival rate of their sex- and birth year-matched peers from the general population. Deviation from the reference line represents better or worse survival (patient survival).

Table 2.

Associations among diagnosis for hip surgery, comorbidities, socioeconomics, and surgical factors with patient survival

Discussion

Insertion of a THA aims to relieve pain, increase function, and improve quality of life, goals that are typically achieved. In addition, we observed that patients who underwent THA in Sweden had a slightly greater life expectancy for the first 10 years after THA compared with the general population. We also identified that patients with primary osteoarthritis of the hip had improved relative survival compared with most other preoperative diagnoses. Finally, worse Elixhauser comorbidity index scores, lower levels of achieved education, and being a widow or single all were associated with poorer survival.

This study has some limitations. Because this a register-based study without a control group, there are no data available about mortality in these patient groups should they not have undergone THA. With an observational study design, the possibility of selection bias can never be fully compensated for. Clinicians select certain patients for procedures. However, because the present study is a nationwide observational study from a country with a publicly funded welfare system, considering all hospitals and all patients, and with adjustment for socioeconomic factors, the possibility of a remaining selection bias is limited and is not regarded to reach such a magnitude as to influence our main findings substantially. Transfer bias was unlikely to be a problem in this study, because leakage from the register is virtually nonexistent. The comorbidity calculations are based on diagnosis established as a result of contacts with the secondary care setting (out- and inpatients within hospitals); this will to a certain extent underestimate comorbidity if this comorbidity has been addressed by the general practitioner and this can well be considered assessment bias. We can only claim to be able to provide information on potential associations and we do not intend to prove causality; specifically, one cannot infer from our data that the act of replacing a hip increases life expectancy. In fact, there is considerable reason to think this is not the case; for example, in this study, we found that patients who underwent surgery at older ages had greater survival than their peers. This may simply reflect the fact that for older patients, only patients with otherwise good health status are considered fit for THA. However, it also is possible that by increasing activity levels and helping patients to discontinue nonsteroidal antiinflammatory drugs (as may occur after successful THA), the procedure indeed may contribute to longevity. Future studies will need to evaluate the causality (or lack of causality) of the findings made here.

The result of the present study, based on a nationwide longitudinally maintained collection of data, largely upholds findings up to 12 years postoperatively as reported by others and corroborates the perception of many clinicians: patients undergoing THA have better survival than that of the general population [21, 25, 33]. A common reasoning behind this is patient selection. Generally to have hip arthroplasty, the patient should be in good health or will undergo some measures to improve their health status as part of their perioperative workup for surgery. A similar reasoning most likely explains the apparent discrepancy between our findings and previous work from the SHAR [14]. Interestingly, Gordon et al. found that patients alive 8 years after their primary operation had a lower survival rate than the reference population [14]. There are some important methodological differences between the studies, but perhaps the major difference is the study population. The population studied by Gordon and collaborators was older and spanned a different time period (1992-2005). We found that the year of the primary operation was associated with better survival; the more recent the operation, the better the survival rate. The assumption of clinicians and care workers is that improvements in surgical practices and perioperative care have caused mortality from hip arthroplasty to decline. The increased relative survival will likely have some consequences on the lifetime risk for revision surgery.

Patients with primary osteoarthritis who underwent THA had a higher survival rate than patients with hip disease resulting from most other diagnoses. The survival rate in the group with primary osteoarthritis was also higher than in the general population during the time period for our study. The results were similar to that of Lie et al. with the exception of sequelae of childhood hip disorders [21]; that study found that those patients had better survival than the corresponding Norwegian population up to 10 years after surgery. However, their data date back to 1987 until 1998, whereas the present study contains data between 1999 and 2012. Thus, it is possible our data contain patients who, for reasons of surgical complexity, would not have been considered for arthroplasty before 1999. Although Norway and Sweden are culturally and socioeconomically similar, there are considerable differences in patient survival rates [23]. The increased mortality risk in patients with inflammatory arthritis has been previously described whether or not they have received a joint replacement and is consistent with our findings [27, 28]. We found that having lower completed levels of education and being widowed or single were associated with an adverse effect on survival; the effect of socioeconomic factors on survival has been long recognized [19]. A recent study concluded that mortality, attributable to lower levels of completed education, is comparable in magnitude to mortality attributable to individuals being current rather than former smokers and the association between education and mortality could be causal [16]. Likewise, it is well known that widowers have higher mortality rates than married people [42] and that married people have a longer life expectancy compared with unmarried persons [15, 34].

As might be predicted, the higher a patient scored on the Elixhauser comorbidity index, the stronger the association with mortality became. The Elixhauser comorbidity index ranks highest among different comorbidity indices that have the ability to predict long-term mortality [41]. Of the studied fixation techniques, patients with cemented, hybrid, and reverse hybrid implants had similar survival rates. The increased mortality rate in the early postoperative period in the patients receiving a cemented implant has once again been demonstrated but is reversed in our series between 50 and 60 days (Fig. 3A, inset). Patients with uncemented prostheses had better survival. This finding was corroborated by McMinn and collaborators [26], but it requires some caution because the mean age in patients who receive cemented prostheses usually is higher (in our study, the mean age of patients receiving a cemented implant was 70 years), whereas the mean age of patients with uncemented prostheses is considerably lower (in our study, the mean age of this group was 55 years at the time of their primary operation). We adjusted for age, but such differences cannot be easily and completely adjusted away, because of the excessive differences in mean age at the time of surgery. If only the relative survival graph is considered, it would be tempting to think that cement has a protective effect after the initial period of excess mortality, but this is more likely to be the result of an increased effect of patient selection in the older age group. An alternative explanation that well could be considered is the smaller incidence of periprosthetic fracture in the cemented group in the postoperative period avoiding the increased mortality of further surgical intervention. This hypothesis will be the subject of further research.

In general terms, elective hip arthroplasty not only provides improvement in health-related quality of life, but is also associated with improved relative survival during the first decade after the operation. This has health economic implications and might cause a shift in the understanding of the lifetime revision risk. Further research in the long-term effects of the increase of the relative survival, the effect on the patient, the implications for healthcare providers, and how we can possibly further modify and improve relative survival is indicated. The effect of fixation technique on mortality will also need to be further scrutinized.