The primary indications for total hip arthroplasty are pain and disability. Although some authors have attempted to define explicit criteria for the appropriateness of total hip arthroplasty1, the subjective nature of the indications will always require patients to weigh the risks and benefits on the basis of their own values. If the results of operative treatment change with the stage of the condition, this factor needs to be taken into account in the decision regarding surgery. For example, if a patient who is otherwise undecided about surgery could expect a better result if he or she underwent a total hip arthroplasty now rather than at some point in the future when his or her clinical condition has worsened, that individual might reasonably choose to undergo the procedure sooner.
The literature is unclear with regard to the effect of a patient's preoperative status on the outcome of total hip arthroplasty. Braeken et al. 2 found that patients with more pain preoperatively reported greater pain between six and twelve months postoperatively on the Rand scale but not on the WOMAC (Western Ontario and McMaster Universities Osteoarthritis Index). MacWilliam et al. 3 found that patients who had a worse preoperative status in terms of function and pain also had a worse status at six months postoperatively compared with more functional patients, but those authors did not further quantify this finding. Finally, Fortin et al. 4 found that patients who had worse preoperative scores on the SF-36 physical function scale and the WOMAC scale had worse scores at six months postoperatively, but the one-year outcomes were not considered.
The purpose of the present study was to examine the relationship between baseline pain and functional status and the outcomes of total hip arthroplasty in a Medicare population while controlling for other diseases that may cause functional pain and disability. This information may be important for patient decision-making regarding total hip arthroplasty.
The data for this study were collected as part of a larger project, the Post Hospitalization Outcome Study (PHOS), the primary purpose of which was to use Medicare administrative data and direct follow-up to develop a method for assessing outcomes for Medicare beneficiaries. Patients who were potentially eligible for the study were identified by the Medicare fiscal intermediary. All elderly Medicare beneficiaries who were sixty-five years old or more, who were not enrolled in a health-maintenance organization, who were discharged with a claim that included the procedure code for total hip replacement (85.51), and who had osteoarthritis were potentially eligible. Patients with hip fracture, bilateral replacement with an interval of one month or less between procedures, or other hip diseases were excluded ( Table I ).
Patients were identified through twelve fiscal intermediaries that historically processed the greatest number of bills for hip replacement. The states that were covered included Texas, California, New York, Florida, Ohio, Pennsylvania, Illinois, Mississippi, North Carolina, Minnesota, Wisconsin, and Indiana. The rationale was to obtain a nationally representative, adequately sized sample. Patients were identified between October 1994 and June 1995.
The subjects were surveyed by telephone within two months after discharge from the hospital, at four to six months after discharge (the time was randomly assigned to test the effect of the duration), and at one year after discharge. In the first of the three surveys, patients were asked about their preoperative status. If the patient could not complete the interview for medical reasons, a proxy completed the interview instead. The patients' hospital records were abstracted with use of a tool especially designed for this project according to guidelines established during the design phase of the project by the clinical experts. Consent was obtained from the subjects at the time of the survey. Approval was obtained from the University of Minnesota internal review board.
The fiscal intermediaries identified 2379 patients who were possibly eligible for the study. Of these, 1640 proved to be eligible for the study; the remainder were not eligible because of missing or inaccurate information, because they were too young, or because their records were not received and processed in time. Of the 1640 eligible patients, 1120 (68.3%) completed the baseline survey and were included in the analysis. Of the 520 eligible patients who did not complete the baseline survey, 155 (30%) could not be located, 134 (26%) were not surveyed within the time-frame of the protocol, 130 (25%) completed the survey too late to be included, fifty-nine (11%) refused, twenty-six (5%) had no available proxy, thirteen (2.5%) had died, and three (0.6%) had not undergone total hip arthroplasty. Of the 1120 patients who completed the baseline survey, 1070 (96%) completed the midpoint survey and 1046 (93%) completed the survey at one year. The medical records of 1066 (95%) of the 1120 patients who completed the baseline survey were abstracted. The other medical records could not be obtained.
Measures
The age and gender of each patient were obtained from the fiscal intermediaries. Comorbid diagnoses that were collected from the medical records included those in the Charlson index 5 and other diagnoses that were thought to be potentially relevant to the outcome of hip replacement surgery ( Table II ).
During the surveys, the patients were asked about their activity level, the presence and severity of pain with walking, whether they needed assistance with walking, how far they could walk, and whether they could perform International Activities of Daily Living (IADLs). A global "functional index," which reflected the number of disabilities that each patient had, was generated by counting whether the patient had severe pain with walking, participated in moderate activity less than daily, needed assistance with walking, needed assistance with grocery shopping, or needed assistance with housework. Moderate activity was defined as "taking a walk, doing light housework, or any other activity you consider moderate" and was distinct from vigorous activity, which was defined as "tennis, running, lifting heavy objects, bicycling."
Analysis
The outcome and baseline measures were treated as dichotomous measures, ordered measures, or continuous measures, depending on the analysis. When treated as continuous measures, the scales were transformed to a range of 0 to 100 for ease of presentation, with 100 representing better status. For this transformation, it was assumed that the levels represented equal intervals on an interval scale.
Unadjusted comparisons between the groups were performed with chi square tests, t tests, and ordinary least squares as appropriate. Adjustment was done with use of logistic regression analysis and ordinary least squares. Statistical analysis was performed with use of STATA software (version 6; STATA, College Station, Texas).
Our findings give further weight to those of previous studies 2-4 , which have suggested that a worse preoperative functional status predicts worse outcome in terms of both function and pain after total hip arthroplasty. The results of the present study also show that the difference persists for at least twelve months following surgery and is independent of other diagnoses that might be expected to cause pain and reduced function, including arthritis in other joints, pulmonary disease, and cardiac disease. The most striking differences were noted for patients who needed assistance with walking, housework, and grocery shopping. It is unclear why the need for assistance had the largest effect. It is possible that the results of surgery are not as good once a patient's condition has deteriorated to the level that necessitates assistance or that once the patient has crossed the threshold of requiring assistance it is difficult to get along without assistance, even if the functional status has improved and pain has decreased. Either way, the outcome is less desirable if the patient requires assistance preoperatively.
Our findings also support the observation 3 that although those who are worse off preoperatively do not achieve, on the average, as good an outcome, they derive the most benefit from surgery. This effect was very striking in terms of both function and pain. Depending on the scale, patients with the lowest levels of preoperative function gained, on the average, between 65 and 89 points (on a 0 to 100-point scale) whereas those with the highest levels of function had a net loss ( Fig. 2 ). Although the most severely affected patients derive the greatest absolute benefit from operative treatment, this does not mean that it is better to wait until an individual has reached this level of disability before an operation is performed.
Determining when to consider a total hip arthroplasty for the treatment of arthritis is difficult. Ultimately, this question needs to be answered by the individual patient with the assistance of his or her physician. Each patient needs to consider his or her own clinical status, social support systems, and values. Additionally, the results of the present study suggest that one factor that some patients should consider is the possibility that the final result of surgery (that is, the degree of disability that will remain after full recovery) may be worse if the procedure is put off to a time when the preoperative functional status is worse. For example, it may be better for a patient to choose to undergo surgery before he or she begins to require the assistance of others because he or she will be less likely to require assistance after surgery.
One reasonable question is how one might apply these findings in the clinical setting. The course of hip arthritis is not such that the physician can reliably tell an individual patient, "you should undergo arthroplasty today because in X months you will be more disabled and have an inferior result." On the other hand, there are patients who, after weighing the risks and benefits of total hip replacement, are truly undecided. These patients may benefit from the knowledge that there may be a price to be paid for waiting to have surgery.
The observation that individuals who are worse off at baseline derive the greatest net benefit has little importance for an individual patient who is trying to decide whether to undergo a total hip arthroplasty. If, at a given point in time, a patient believes that the overall benefit of total hip arthroplasty outweighs the risks, then delaying the procedure until the benefit is even greater makes no sense.
Like most observational studies, this one is limited by the possibility that the differences that were seen were really associated with unobserved variables. Specifically, it is possible that we have not adequately controlled for other causes of disability and, therefore, that patients who were more disabled did worse because they had disabilities that were not addressed by total hip arthroplasty. Although this is possible, we did control for the major diseases that would be expected to cause pain and disability in this population and still observed more than a twenty-percentage-point difference in the proportion of patients who needed assistance in a number of dimensions postoperatively. It is possible that better control could make this difference smaller, but it seems unlikely to disappear.
Another possible limitation of this study is that the preoperative status was ascertained as long as two months after the operation, but this factor seems unlikely to have biased the results. If anything, one might expect that the measurement of preoperative status would be less reliable, which would decrease, rather than falsely increase, the strength of the observed associations.
In this nationally representative population of Medicare beneficiaries, we observed that some patients had a suboptimal result because they underwent surgery after they had become substantially disabled. There are many possible reasons that a patient may undergo surgery later in the course of their disease, including personal values about the risks and benefits of surgery, differing social support systems, and the unpredictability of the course of the disease. What is important is that patients considering total hip arthroplasty understand not only the risks and benefits at the present time but also how those risks and benefits may change if the consideration of surgery is put off to another day.