Determinants of demand for total hip and knee arthroplasty: a systematic literature review

Abstract

Background: Documented age, gender, race and socio-economic disparities in total joint arthroplasty (TJA), suggest that those who need the surgery may not receive it, and present a challenge to explain the causes of unmet need. It is not clear whether doctors limit treatment opportunities to patients, nor is it known the effect that patient beliefs and expectations about the operation, including their paid work status and retirement plans, have on the decision to undergo TJA. Identifying socio-economic and other determinants of demand would inform the design of effective and efficient health policy. This review was conducted to identify the factors that lead patients in need to undergo TJA.

Methods: An electronic search of the Embase and Medline (Ovid) bibliographic databases conducted in September 2011 identified studies in the English language that reported on factors driving patients in need of hip or knee replacement to undergo surgery. The review included reports of elective surgery rates in eligible patients or, controlling for disease severity, in general subjects, and stated clinical experts' and patients' opinions on suitability for or willingness to undergo TJA. Quantitative and qualitative studies were reviewed, but quantitative studies involving fewer than 20 subjects were excluded. The quality of individual studies was assessed on the basis of study design (i.e., prospective versus retrospective), reporting of attrition, adjustment for and report of confounding effects, and reported measures of need (self-reported versus doctor-assessed). Reported estimates of effect on the probability of surgery from analyses adjusting for confounders were summarised in narrative form and synthesised in odds ratio (OR) forest plots for individual determinants.

Results: The review included 26 quantitative studies-23 on individuals' decisions or views on having the operation and three about health professionals' opinions-and 10 qualitative studies. Ethnic and racial disparities in TJA use are associated with socio-economic access factors and expectations about the process and outcomes of surgery. In the United States, health insurance coverage affects demand, including that from the Medicare population, for whom having supplemental Medicaid coverage increases the likelihood of undergoing TJA. Patients with post-secondary education are more likely to demand hip or knee surgery than those without it (range of OR 0.87-2.38). Women are as willing to undergo surgery as men, but they are less likely to be offered surgery by specialists than men with the same need. There is considerable variation in patient demand with age, with distinct patterns for hip and knee. Paid employment appears to increase the chances of undergoing surgery, but no study was found that investigated the relationship between retirement plans and demand for TJA. There is evidence of substantial geographical variation in access to joint replacement within the territory covered by a public national health system, which is unlikely to be explained by differences in preference or unmeasured need alone. The literature tends to focus on associations, rather than testing of causal relationships, and is insufficient to assess the relative importance of determinants.

Conclusions: Patients' use of hip and knee replacement is a function of their socio-economic circumstances, which reinforce disparities by gender and race originating in the doctor-patient interaction. Willingness to undergo surgery declines steeply after the age of retirement, at the time some eligible patients may lower their expectations of health status achievement. There is some evidence that paid employment independently increases the likelihood of operation. The relative contribution of variations in surgical decision making to differential access across regions within countries deserves further research that controls for clinical need and patient lifestyle preferences, including retirement decisions. Evidence on this question will become increasingly relevant for service planning and policy design in societies with ageing populations.

Figure 1

Process of identification of demand studies.

Figure 2

Ratios: a) age ≥ 60 to age 50–59 surgery rates (left) and b) propensity for surgery at age > 62 relative to age ≤62 rate (right). * Rates in public hospitals in England, adjusting for gender, ward ethnic mix and deprivation, distance to and characteristics of hospitals [28]. ** Rate of receipt of primary surgery after five years, based on population samples from two areas of Ontario, Canada. Relative to age ≤ 62. Adjusting for WOMAC, SF-36 General Health, Willingness to undergo surgery [26].

Figure 3

Ratio of male to female surgery rate. *U.S. population-based study age ≥ 47. Patients with an arthritis-related visit to the doctor in the past two years at baseline. Adjusted for age, gender, race, comorbidities, functional limitations, income, wealth, insurance type, employment, and BMI ≥ 25 [21]. ~ Rate over 18 months, from south of England study of patients aged ≥65. Unadjusted for confounders [48]. #Surgical rates in public hospitals, population of England. One year incidence, adjusted for socio-economic and ethnicity mix of ward of residence [28]. ^Respondents reporting an arthritis-related visit at baseline. Adjusting for demographics, health need and economic access (including health insurance, wealth and education) [38].

Figure 4

Ratio of surgical rate and willingness to undergo surgery by race minority group to white. Dunlop: Adjusting for demographics, health need and economic access (including health insurance, wealth and education). Black/Hispanic refers to all respondents; Black/Hispanic^ refers to respondents reporting an arthritis-related visit at baseline [29]. Steel: Adjusting for covariates age, gender, wealth, employment, BMI ≥ 30, seen doctor ≥2 times in last 2 years, education, comorbidity, grandchild care, difficulty walking 1 block or more, married/cohabiting. Other category is defined as non-black and non-white [38]. Ibrahim: AA African-American sample adjusting for age, level of education, annual income, radiologic severity of disease, WOMAC, geriatric depression score; AA* Adjusting for familiarity with surgery in addition to covariates for AA[40]; AA** Adjusting for familiarity and expectations in addition to covariates for AA [40]. Suarez-Almazor: Adjusted for age, gender, years of education, trust in physician, perception of efficacy, perception of risk, WOMAC, relative/friend with TKA. AA refers to African American sample [44].

Figure 5

Ratio of surgical rate with post-secondary education to rate with lower education level. Hanchate: Relative to less than high school education. U.S. population-based study age ≥ 47. Adjusted for age, gender, race, comorbidities, functional limitations, income, wealth, insurance type, employment, BMI ≥ 25. *Refers to respondents with an arthritis-related visit to the doctor in the past two years at baseline; other result refers to all respondents [21]. Hawker: Relative to less than high school education. Sample from two areas in Ontario, Canada. Adjusting for WOMAC, SF-36 General Health, age [26]. Steel: Relative to less than college education. U.S. population-based study age ≥ 60. Adjusting for covariates age, gender, wealth, employment, BMI ≥ 30, seen doctor ≥2 times in past 2 years, comorbidity [38].

Figure 6

Ratio of surgical rate overweight (BMI ≥ 25) or obese (BMI ≥ 30) to rate with a lower BMI. ^U.S. population-based study age ≥ 47. Patients with an arthritis-related visit to the doctor in the past two years at baseline. Adjusted for age, gender, race, comorbidities, functional limitations, income, wealth, insurance type, employment, and education [21]. *Arthritis subsample. Other details as in Footnote ^. **Three-year rate. Adjusted for age, morning stiffness, Kellgren-Lawrence ≥ 2, worst pain location, function with decreased active hip motion [19]. ***Six-year rate. Other details as in Footnote **. ~U.S. population-based study age ≥ 60. Adjusting for covariates age, sex, wealth, employment, seen doctor ≥2 times in past 2 years, education, and comorbidity [38].