SEARCH

SEARCH BY CITATION

Keywords:

  • Anthroplasty;
  • Mortality;
  • Centenarian;
  • Nonagenarian;
  • Aging

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. REFERENCES

Objective

The number of individuals ages ≥100 years in the US is expected to increase considerably in the future along with the need for arthroplasties. This report focuses on the poorly studied epidemiology and mortality outcomes of arthroplasty among these individuals.

Methods

We describe the epidemiology of knee and hip arthroplasties among centenarians using data from a large hospital discharge database in the US (the Nationwide Inpatient Sample) during the period 1993 through 2002. We used nonagenarians as the comparison group with adjustment for differences in the prevalence of congestive heart failure, neurologic diseases such as dementia and stroke, renal and hepatic diseases, obesity, anemia, malignancy, coagulopathy, and depression and other psychiatric illnesses. Cox regression models were used to study the mortality outcomes following arthroplasty.

Results

Overall, there were 679 hip arthroplasties and 7 knee arthroplasties among centenarians in this database. The corresponding figures for nonagenarians were 33,975 and 2,050, respectively. A vast majority (83%) of hip arthroplasty recipients were women. Risk-adjusted mortality estimates following arthroplasty for centenarians were higher than for nonagenarians (hazard ratio 1.46, 95% confidence interval 1.10–1.95). However, this was similar to differences in overall in-hospital mortality (hazard ratio 1.36, 95% confidence interval 1.32–1.40) between these 2 age categories.

Conclusion

In the US population, hip and knee arthroplasty are very rarely performed among centenarians. Our in-hospital mortality data suggest that arthroplasties should not be denied to centenarians solely because of short-term postoperative life expectancy estimates.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. REFERENCES

Aging is inevitable, as are problems with organ failure such as osteoarthritis and consequences of frailty such as osteoporotic hip fractures. From a musculoskeletal specialist's perspective, these are 2 major causes of disability among the older age group.

Centenarians are individuals who have avoided or delayed various diseases that lead to death in the first 100 years of life and are a living example of successful aging (1). Even though ∼40% of centenarians are functionally independent (1), they are among those at the highest risk for both disabling arthritis and osteoporotic fractures. Arthroplasty is effective but is a treatment of last resort for these musculoskeletal problems. However, with increasing age, the safety and desirability of performing such expensive and invasive procedures may be questioned for 2 reasons. The first reason is the utilitarian view that finite health care resources should be spent on individuals in which the potential benefits are highest. Thus, older individuals with fewer potential remaining years of life are less “deserving” of expensive “investment” than younger persons. The second reason is biomedical. Arthroplasty may be too hazardous for the older patient due to frailty, defined as a combination of inability to deal with physical and psychological stress and physical impairment. With respect to arthroplasties, however, there are comparatively few data that inform the above debate. Due to the relative rarity of centenarians and arthroplasties among them, such data are more likely to emerge from very large, nationally representative databases than from single-site or multicenter studies. In this report, we provide such data on the utilization and short-term outcomes of arthroplasties among the oldest age group in the US.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. REFERENCES

The database.

We used 10-year data from the Healthcare Cost and Utilization Project Nationwide Inpatient Sample (NIS) for the years 1993 through 2002. The NIS is the largest hospital discharge data set in the world. These data are collected from all discharges from a 20% sample of US community hospitals. Collectively they contain records of approximately 7 million hospital discharges per year. Each record in the NIS contains data from the discharge summary that was abstracted and entered into a computer by trained coders and data entry operators. The data items include age, sex, nature of hospitalization (emergent versus routine), up to 15 diagnoses, up to 15 procedures performed during hospitalization, length of hospital stay, and vital status at the end of hospitalization. The NIS data have been well described and have been utilized for health outcomes research on joint replacements (2, 3).

Subject selection and outcomes.

All hospitalizations where primary hip and knee arthroplasty among centenarians was performed were abstracted from the NIS (International Classification of Diseases, Ninth Revision [ICD-9] procedure codes 81.51–81.52 for hip arthroplasties and 81.54 for knee arthroplasties). Indication for surgery was identified using the ICD-9 clinical modification code appearing as the primary diagnosis for hospitalization. Individuals were categorized as centenarians if they were at least 100 years old at the time of the index hospitalization. Indication for arthroplasty was classified as elective or nonelective. The comparison group was hospitalizations of all individuals ages 90–99 years (nonagenarians). In-hospital mortality rate was the outcome studied. Because of patient privacy issues, individual patient-identifying variables were not available for analysis and, therefore, vital status after hospital discharge could not be determined.

Statistical analyses.

For calculating incidence rates of arthroplasties, we used the 1995 intercensal estimate of the number of centenarians (n = 54,000) as the denominator (4). In-hospital mortality rate was calculated as the number of deaths per person-year of hospitalization. The latter was calculated as the cumulative number of hospitalized days for each of the subgroups studied. In addition, the mortality statistics were examined as a proportion of hospitalizations that resulted in death.

There are no easy controls for studying the impact of age on orthopedic surgery outcomes among centenarians. We therefore used nonagenarians as a comparison group. For studying the relative mortality risk of centenarians compared with nonagenarians, we used Cox proportional hazards regression models (5). In these models, person-years of hospitalization and risk of in-hospital death were jointly modeled after adjusting for sex and the following comorbidities: congestive heart failure, neurologic diseases such as dementia and stroke, renal and hepatic diseases, obesity, anemia, malignancy, coagulopathy, and depression and other psychiatric illnesses. Other diseases such as coronary artery disease were not used as a covariate due to their very high frequency in this population.

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. REFERENCES

Overall, there were 57 million hospitalization records in the NIS during the period 1993 to 2002. Of these, there were 41,335 hospitalizations for centenarians. There were 1.6 million hospitalizations in the nonagenarian age group, of which 150,000 died in the hospital.

Arthroplasty among centenarians and nonagenarians.

Within centenarians, we identified 679 primary total hip arthroplasties and 7 primary total knee arthroplasties (Table 1). In age- and sex-adjusted logistic regression models, no significant time trends were observable in the likelihood for hip/knee replacements over the study period. Accounting for the 20% sampling, and assuming a 1995 base centenarian population of 54,000, the annual rate of hip arthroplasties was estimated to be 63 per 10,000 centenarians per annum, whereas that of knee arthroplasty was 5 per 10,000 per annum. The corresponding rate estimates for nonagenarians (base population in 1995: 1.25 million) were 136 hip arthroplasties per 10,000 per annum and 8 knee arthroplasties per 10,000 per annum. In separate logistic regression models for centenarians and nonagenarians, sex-adjusted time trends for both arthroplasties in both age groups were statistically not significant but the direction of trend was upward for knee arthroplasties and downward for hip arthroplasties.

Table 1. Frequency of hip and knee arthroplasties in the Nationwide Inpatient Sample 1993–2002
 HipKnee
NonelectiveElectiveTotalNonelectiveElectiveTotal
Nonagenarians30,3193,65633,9754881,5622,050
Centenarians64633679437
Total30,9653,68934,6544921,5652,057

The median age among centenarians who underwent hip replacement was 101 years (range 100–109 years). The mean ± SD length of stay was 7.5 ± 6.9 days. The elective/nonelective nature of the surgery was not available in 2 instances of knee arthroplasties and 91 cases of hip arthroplasties. Overall, there were 686 arthroplasties, and 564 (83%) of the hip arthroplasty recipients and 3 of 7 knee replacement recipients were women. Thirty-three (5%) hip arthroplasties and 3 of the 5 knee arthroplasties were performed during elective hospitalizations. In comparison, the mean age of nonagenarians undergoing hip arthroplasty (n = 33,975) was 92 ± 2 years, and the mean ± SD length of stay was 7.4 ± 5 days. Among nonagenarians, 12% of hip replacements and 86% of knee replacements were elective. Fractures constituted 94% of nonelective hip arthroplasties among nonagenarians and centenarians. Half of the elective knee replacements were performed for osteoarthritis whereas elective hip arthroplasties were performed primarily for aseptic necrosis and malunion.

Among centenarians, osteoarthritis was the indication for knee replacement in 5 of 7 cases, and these were performed as elective procedures. The remaining knee arthroplasties were performed as emergency cases, and the main cause attributed was fractures. Nonunion or malunion of fractures otherwise treated (1.6%) and aseptic necrosis (0.4%) constituted the remaining cases. Osteoarthritis was among the least common reasons for this surgery (0.2%). Only 5% of hip arthroplasties (n = 33) were performed on an elective basis, and the rest were performed emergently for hip fracture.

Arthroplasty outcomes among centenarians compared with nonagenarians.

The in-hospital mortality rates among centenarians and nonagenarians are shown in Table 2. The overall mortality rate for centenarians was 6.1 deaths per person-year of hospitalization (95% confidence interval [95% CI] 5.9–6.3). The corresponding figure for nonagenarians was 4.5 (95% CI 4.47–4.52). After adjusting for sex and comorbidities, centenarians were at a higher risk for in-hospital mortality. In multivariable Cox regression models limited to hospitalizations for hip arthroplasties alone, mortality risk was higher among centenarians who underwent hip arthroplasty compared with nonagenarians who underwent a similar procedure (hazard ratio 1.45, 95% CI 1.10–1.95). When the analyses were performed using records of all centenarians, hospitalization for hip replacement was associated with a lower risk of death. The adjusted hazard ratio was 0.85 (95% CI 0.75–0.66) for hospitalizations for hip arthroplasty compared with all other causes of hospitalization combined. This lower risk for hip arthroplasty was also found when analyses were confined to nonagenarians (adjusted hazard ratio 0.51, 95% CI 0.49–0.54).

Table 2. In-hospital mortality rates and adjusted mortality risk: a comparison of centenarians and nonagenarians*
 NonagenariansCentenarians
No. of hospitalizationsNo. of deaths (percentage died in hospital)Mortality rate (95% CI)No. of hospitalizationsNo. of deaths (percentage died in hospital)Mortality rate (95% CI)Risk-adjusted hazards ratio (95% CI)
  • *

    95% CI = 95% confidence interval; NA = not applicable.

  • Mortality risk for in-hospital death for centenarians compared with nonagenarians was calculated by Cox regression models where the covariates adjusted for were congestive heart failure, anemia, renal failure, diabetes, chronic lung diseases, and hypertension.

All causes1.6 million149,183 (9.1)4.50 (4.47–4.52)41,3355,033 (12.2)6.1 (5.9–6.3)1.36 (1.32–1.40)
Knee arthroplasty2,05226 (1.3)0.8 (0.6–1.2)70.0 (0.0)0.0NA
Hip arthroplasty33,8511,597 (4.7)2.3 (2.2–2.4)67845 (6.6)3.2 (2.4–4.3)1.45 (1.10–1.95)

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. REFERENCES

There have been few studies on arthroplasty among nonagenarians and centenarians. Because of the rarity of centenarians and arthroplasty performed on them, previous studies assessing age and arthroplasty grouped all such subjects with those in younger age categories (6–9). Ours is the largest series of primary total hip and knee arthroplasty among centenarians in the published literature. Data from the NIS presented here demonstrate that elective knee and hip arthroplasties are very infrequent medical procedures in this age group.

The postoperative mortality of centenarians, as a proportion of the number of hospitalizations, was lower than the proportion of all-cause mortality for these individuals. This finding probably reflects the selection bias due to channeling of low-risk candidates for arthroplasty. This mortality risk following hip replacement (12%) is several magnitudes higher than the rates (0.26%) reported for all elective knee and hip arthroplasties in the general population (10).

In this study, knee arthroplasty was less frequent than hip arthroplasty. Half of the elective knee replacements were performed for osteoarthritis whereas elective hip arthroplasties were performed primarily for aseptic necrosis and malunion. This relatively low frequency of elective surgery might be due to physician and patient judgment that these individuals are at high risk for poor outcomes and that the risk is not offset by the perceived benefit in light of the relatively short life expectancy (11). The most common indication for hip arthroplasty among centenarians was hip fracture in contrast to osteoarthritis among octogenarians and nonagenarians (9).

Utilitarians such as Callahan (12) have argued that the government should pay only for palliative measures beyond the age of 70 or 80 and not for life-extending measures. By that standard, one may argue that arthroplasty, even though reflecting a one-time expense, is cost effective by increasing patient independence and utilization of ambulatory health care facilities and perhaps even providing cardiovascular health benefits. Indeed medical expenses in the last year of life account for a larger share of Medicare expenditures than in earlier years (13).

Although frailty is known to increase with age, there have been arguments that with better medical care such frailty is being compressed into the extremes of age (14). Centenarians live as long as they do by delaying or even avoiding many age-related diseases. Even among those who live with such diseases for a long time, many appear to do so with better functional status than do younger persons who do not achieve extreme old age. In other words, centenarians are those who have compressed frailty toward the very last few years of their life (15). This could mean that these individuals are at greatest risk for poor surgical outcomes. Alternatively, one could argue that arthroplasty helps centenarians to further postpone and compress the period of morbidity that precedes death.

There are scant data to study the probability of centenarians surviving the immediate postoperative intensive care period after undergoing major surgery such as arthroplasty. A report from the Cedar Sinai Medical Center, Los Angeles demonstrated that surgical intensive care outcomes of 140 nonagenarians were not different from younger patients after adjusting for severity of illness (16). Among centenarians only 2 of 9 hospitalizations ended in the death of the individual (17).

Will centenarians and nonagenarians be able to utilize the benefits from a new hip or knee? The Heidelberg Centenarian Study demonstrated that approximately half of the centenarian population showed moderate to severe cognitive impairment, whereas one-quarter were found to be cognitively intact (18). Furthermore, among those individuals, the age-related cognitive decline decelerated with time, suggesting that centenarians can sustain their mental functioning over time. Another study performed with a cohort of nonagenarians showed that among those surgically treated for hip fractures, being 90 years of age and older did not increase the chance of having a postoperative complication, being placed in a skilled nursing facility at discharge or at 1-year followup, or recovering of prefracture independence in instrumental activities of daily living (19).

Our study's strengths are the nationally representative sample, relatively large number of centenarians, and index procedures and thus robust estimates of in-hospital mortality. Our results are less likely to be biased than single-center reports. In contrast, there are several limitations. Most important is the lack of clinical information about the patient's risk profile, surgical technique, and type of surgeon. Second, in-hospital perioperative mortality may not reflect the overall surgical mortality risk. We did not have out-of-hospital followup data that would have permitted calculation of 28-day, 6-month, and 1-year mortality rates. Another issue is the small number of knee arthroplasties, which makes it difficult to generate estimates of procedure rates among centenarians. From a statistical perspective, the relatively small numbers of procedures compared with the underlying population among centenarians result in a wide confidence interval and therefore the population rates that we have provided should be interpreted as the best estimate and not the actual number of arthroplasties. Last, we relied on the accuracy of the administrative database to assess indications for surgery. Therefore, coding errors could be present in the cause of surgery ICD-9 codes. Age inflation exists in most age databases that rely on census data for their centenarian statistics, including in the US, where a national birth registration system did not exist at the time centenarians were born (20). This is also a potential limitation of our data.

According to the US Department of Census, the number of centenarians could cross the 4 million mark by the year 2050 (21). Over the past decade, the utilization of arthroplasties has been increasing in the US (22), and this trend is likely to continue. Due to the impact of hip fractures, the main indication for hip arthroplasty among centenarians can be considerable. This study provides data that suggest arthroplasty need not be denied to centenarians solely on account of age and the concern of high in-hospital mortality risk.

AUTHOR CONTRIBUTIONS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. REFERENCES

Dr. Krishnan had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study design. Krishnan, Fries, Kwoh.

Acquisition of data. Krishnan.

Analysis and interpretation of data. Krishnan, Fries, Kwoh.

Manuscript preparation. Krishnan, Fries, Kwoh.

Statistical analysis. Krishnan.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. REFERENCES