View the most recent version.

Findings

Joint (hip and knee) replacement can provide substantial relief to people suffering from pain and limited mobility. In Canada, approximately 23,000 hip replacements and 38,400 knee replacements were conducted in 2006/2007.¹ The rate at which these procedures were performed more than doubled between 1995/1996 and 2005/2006, with even sharper increases between 2004/2005 and 2006/2007.² The rising rate is partially a reflection of an aging population; the recent acceleration is likely related to the identification of joint replacement among the five priority areas selected for meaningful reductions in waiting times.³

Accumulating evidence points to the health benefits of joint replacement for osteoarthritis, including reduced pain and greater mobility, which improve health-related quality of life.^4-9 But despite generally positive results, some patients do not appear to benefit from these procedures.¹⁰ Recent reviews by Jones et al. indicated that 15% to 30% of arthroplasty patients reported little or no improvement in health-related quality of life after surgery.^11,12 However, the generalizability of most outcome studies is limited, as they were based on selected samples representing specific geographic regions, institutions, clinical sites, and/or providers.

The evidence is less clear about the effectiveness of hip replacement for hip fracture patients. Considerable disagreement remains about the best course of treatment, depending on factors such as age, type of fracture and condition of the hip.^13-15 While surgery is almost always indicated for such patients, the indications for type of surgery are less clear for some subtypes of hip fracture.¹⁵ Some studies report higher rates of infection and mortality after hip replacement, compared with alternative procedures such as internal fixation.^16-18 Other studies report lower rates of re-operation and comparable hip function and health-related quality of life in the long term.^18,19

A population approach to health outcomes yields information about the results of care obtained in various settings, representing a wide range of patients, providers and health care institutions. Most previous research in this area has relied either on survey data, which offer only limited information about health care services received, or on administrative data, which often lack information about health outcomes and about patient characteristics that may explain why some fare better than others.

This study takes an innovative, population-based approach to the analysis of health outcomes using linked survey and administrative data. Responses to the 2000/2001 Canadian Community Health Survey (CCHS) were linked with administrative data from the Hospital Morbidity Database (HMDB) on the use of inpatient acute-care services. Linkage of these two datasets makes it possible to take advantage of the strengths of each.

The primary objective is to study patients’ health outcomes after hip and knee replacement: specifically, whether those who have these procedures (surgical group) return to the average health status of their peers (comparison group). Combining patient-based information from the CCHS and from the HMDB allows for an investigation of a wide range of factors hypothesized to be associated with outcomes of care, as identified in the Health Outcome Framework developed by Statistics Canada and the Canadian Institute for Health Information.²⁰

The second, more data-driven, objective is to examine the potential of linked data for the analysis of health outcomes of specific surgical interventions. This will provide some policy perspective on gains to be made in future data investments, for example, surveys of patients who have undergone surgical interventions.

Methods

Data source

The data are from the Canadian Community Health Survey (CCHS) and the Hospital Morbidity Database (HMDB). The CCHS is a nationally representative cross-sectional survey that collects information about Canadians’ health status and use of health care. Cycle 1.1 was conducted in 2000/2001 with a sample size of 131,535.²¹ The survey covers approximately 98% of the population aged 15 or older living in private dwellings.

The HMDB is a national administrative database representing all inpatient acute hospital admissions. It contains information on dates of admission and separation, up to sixteen ICD-9 diagnoses identifying the reason(s) for the stay, and up to ten procedure codes (based on ICD-9/-10 codes²²) indicating interventions during the stay.

Study sample

To identify the “surgical group” (those who had jointreplacement surgery), data from cycle 1.1 of the CCHS were linked to HMDB data covering the five years before the survey (1995/1996 to 2000/2001) using probabilistic data linkage techniques based on health insurance number, sex, date of birth and postal code.^23,24 The analyses included only respondents who agreed to have their survey information linked to administrative data. The Statistics Canada Policy Committee approved the linkage. To address potential bias introduced by non-linkers, new survey weights were derived. The analyses excluded CCHS respondents from Quebec, because data provided to Statistics Canada by Quebec for the HMDB have scrambled health insurance numbers, which make it impossible to link administrative records and survey responses.

Hospital stays were included in the analysis only if they were coded with a first surgical intervention indicating hip or knee replacement (Table 1). Some individuals had more than one acute inpatient admission with the relevant procedure codes. In these cases, the hospital event closest to the survey date was retained for analysis; subsequent admissions were dropped. No attempt was made to differentiate revisions from primary replacements; individuals (n=16) who stayed in hospital for these surgeries both before and after their CCHS interview were excluded. As well, hospital stays that occurred within the six months before the CCHS interview were excluded, because in these cases, answers to the survey questions about heath status would reflect the post-operative recovery/rehabilitation period rather than full recovery. The sample was limited to CCHS respondents aged 40 or older because joint replacement at younger ages is rare and generally has different precursors and causes.

Table 1
Procedure and diagnosis codes used to identify surgery groups

The “comparison group” consisted of CCHS respondents aged 40 years or older who had not had joint replacement in the five years before their interview (n=58,667).

Analytical techniques

Univariate analyses and weighted multivariate linear regression were used to compare the health status of individuals who had joint replacement surgery (“surgical group”) with those who did not (“comparison group”), controlling for factors associated with post-operative health status. The same variables were then modelled to identify factors associated with health status among surgical patients diagnosed with osteoarthritis. Small sample sizes prevented similar analyses for the other diagnostic groups. Analyses were conducted with Stata software using the xi: regression procedure. Special linkage weights were developed by Statistics Canada to adjust the linked data for those who did not consent to link and those who could not be linked because the information required for linkage was insufficient.

Variables

Health outcome measure

The primary health outcome measure is the health utility index (HUI), a multidimensional preference-based measure of health status^25,26 that has been used in studies of population health^27-29 and in clinical settings,³⁰ including among joint replacement patients. The HUI has a theoretical range between -0.3 (living in a state worse than death) and 1 (perfect health). It is intended to capture an individual’s functional health status across eight dimensions: vision, hearing, speech, dexterity, cognition, emotion, mobility and pain. The two latter dimensions are particularly relevant for individuals undergoing hip and knee replacement surgery. A difference of 0.03 in the HUI is considered clinically significant.²⁹

Independent variables

The CCHS includes demographic information (age, sex, marital status, province of residence), socio-economic variables (household income, education), and risk factors that are hypothesized to be related to health status (presence of chronic conditions, body mass index, smoking). Education refers to the highest level attained by the respondent: less than secondary graduation; secondary graduation or some postsecondary, and postsecondary graduation. Household income, adjusted for household size, was measured in quintiles.

The CCHS collects information about chronic conditions including arthritis, diabetes, chronic obstructive pulmonary disease, asthma, hypertension, stroke, heart conditions, chronic pain, cancer and depression. Individuals were classified by the number chronic conditions they reported as diagnosed by a health professional and lasting more than six months. Body mass index (BMI) was based on self-reported height and weight (weight in kilograms/height in metres squared). Smoking status was categorized as never smoked, former smoker, or current smoker based on self-reported smoking habits.

The surgical cohort was divided into diagnostic groups according to the reason for joint replacement as indicated by the most responsible diagnosis code on the hospital separation record for the surgical procedure: osteoarthritis, fracture (hip replacements), complications (specific ICD codes indicating complications of a surgical intervention), or other (for example, cancer, rheumatoid arthritis). This classification reflects the hypothesis that post-operative recovery differs depending on the reason for the surgery. Individuals undergoing joint replacement due to fractures, for example, experience a different trajectory of care and outcomes, given that the surgery is in response to an acute event.³¹

Results

Descriptive

A total of 598 individuals had a hip or knee replacement sometime between six months and five years before their CCHS cycle 1.1 interview (Table 2). Osteoarthritis was the most common diagnosis among both hip and knee replacement patients: 29.5% and 40.0%, respectively. Hip fractures accounted for 8.7% of the cohort. Almost equal percentages had a joint replaced with or resulting from complications (10.5%), or with other diagnoses such as cancer or rheumatoid arthritis (11.2%).

Table 2
Distribution of surgery groups, by surgical procedure and diagnosis, respondents aged 40 or older to 2000/2001 Canadian Community Health Survey, Canada excluding Quebec

The surgical group was, on average, older than the comparison group (47.3% versus 10.3% were aged 75 or older) and more likely to be female (63.4% versus 51.6%) and to have co-morbidities (89.7% versus 52.4%) (Table 3).

Table 3
Selected characteristics of surgery and comparison groups, respondents aged 40 or older to 2000/2001 Canadian Community Health Survey, Canada excluding Quebec

Average (unadjusted) health status, measured by the HUI, was 0.615 for the surgical group and 0.844 for the comparison group (Table 3). The differences were mostly in the dimensions of mobility and pain, as shown, for example, on a radar plot for those age 65 to 74 (Figure 1). The pattern was similar for the other age groups and when the fracture group was removed from the analysis (data not shown).

Figure 1
Mean (unadjusted) Health Utility Index scores, by attribute, for surgery and comparison  groups, aged 65 to 74, to 2000/2001 Canadian Community Health Survey, Canada excluding Quebec

Multivariate regression analysis

Overall, the surgical group reported lower functional health than did the comparison group, when the results were adjusted for other covariates hypothesized to be associated with health (Table 4). The results, however, varied by diagnosis. Joint replacement patients with a primary diagnosis of osteoarthritis “regained” more health, reporting 6% (hip replacement) and 9% (knee replacement) less functional health compared with the control group, whereas the hip facture group reported 21% less functional health.

Table 4
Adjusted† difference in Health Utility Index between surgical and comparison groups, by surgical procedure and diagnosis, respondents aged 40 or older to 2000/2001 Canadian Community Health Survey, Canada excluding Quebec

Among joint replacement patients with osteoarthritis, several other factors were significantly associated with post-operative health status (Table 5). Their functional health decreased with each additional chronic condition (13% less). Those who were underweight reported 24% less functional health, compared with “normal” weight individuals. Former smokers reported more functional health (7%), compared with those who never smoked.

Table 5
Linear regression coefficients relating selected characteristics to Health Utility Index, joint replacement patients with osteoarthritis, respondents aged 40 or older to 2000/2001 Canadian Community Health Survey, Canada excluding Quebec

Discussion

This is the first population-based analysis of health outcomes of joint replacement using linked survey and administrative data at the national level in Canada. Unlike studies based solely on administrative health data, the availability of health-related quality of life information (HUI) in the survey data allowed a more direct assessment of health outcomes on a range of patients, in a variety of care settings and providers.

On average, individuals who underwent joint replacement surgery were not restored to a level of functional level compared with a similar population group. As expected, the results varied by type of diagnosis, from 6% (hip replacement) and 9% (knee replacement) lower functional health among those with a diagnosis of osteoarthritis to 21% lower functional health among the hip fracture group. After surgery, patients with fractures do not “regain” health to the same degree as the osteoarthritis group. This finding supports evidence about the outcomes of treatment for hip fractures. Hip fracture has been associated with excess mortality, compared with the general population³² and compared with other hip replacement recipients.³³ As previously observed, the evidence about the effectiveness of joint replacement for hip fracture patients is mixed. Other consequences of hip fractures may adversely affect patients’ health-related quality of life. It is likely that the fracture itself has a negative impact on their health trajectory; for example, the hospital stay itself can result in changes in functional status.^34-36 Fractures among the elderly are as much a cause as a consequence of frailty, representing a closer to terminal event in the process of health decline.^37,38

The linked database made it possible to explore a range of factors associated with health outcomes of joint replacement among a nationally representative population. The results indicate that, among people with osteoarthritis who underwent joint replacement, being underweight and having co-morbid conditions were associated with poorer post-operative health. Although sex, age and marital status also seemed to be associated with poorer health, the results did not attain statistical significance, likely because of the small sample size. These results are consistent with other findings that point to a variety of factors associated with outcomes of joint replacement,^39,40 including co-morbid conditions⁴¹ and lack of social support.⁴² These associations may indicate the expected effectiveness of joint replacement, in terms of health status, for individuals with osteoarthritis.

The better health of former smokers, compared with those who never smoked, was unanticipated. However, former smokers include both recent and long-time quitters, the latter of whom often achieve health status and adopt health care practices similar to those of non-smokers.^43,44 In fact, some evidence suggests that long-time quitters are more likely than non-smokers to believe in the efficacy of modifying other risk factors.⁴⁵ It is possible, then, that former smokers (at least, long-time quitters) have adopted other healthy lifestyles, such as greater physical activity, that improve their overall health.

Limitations

This study has several limitations. First, the sample size is small—the analysis pertains only to joint replacement patients who were respondents to the 2000/2001 CCHS. Subsequent studies may benefit from ongoing efforts at Statistics Canada to link several waves of the CCHS to hospital administrative data. This limitation, however, is counterbalanced by gains in generalizability—the data represent the Canadian population, not a single hospital or a single health insurance provider or even a single province.

Second, because the sample is restricted to the household population, it does not represent outcomes of joint replacement among residents of institutions such as long-term care facilities.

Finally, the study does not directly measure the change in health status before and after surgery. Rather, it compares the post-operative health status of surgical patients to a population comparison group. This approach assumes that the surgery was intended to restore patients to a level of health similar to that of their contemporaries. However, a negative finding does not necessarily signal the absence of a gain in health-related quality of life as a result of the surgery.

Conclusion

This study is a unique application of linked data to the study of health outcomes after a health care intervention, namely, joint replacement. The data allow for a population approach to the assessment of health outcomes, taking into account a range of factors. The outcomes of joint replacement differ depending on the initial diagnosis or reason for the surgery. In particular, patients with osteoarthritis who are underweight or have co-morbid conditions may be susceptible to poorer outcomes. Linked data show promise for studying outcomes of health care interventions, especially interventions that are common and are well-documented in administrative records.