|
|
| |||||||||||||||||||||||||||||||||||||||
Background Canada, which has universal single-payer health insurance, is often criticized for waiting times for surgery that are longer than those in the United States. We compared waiting times for orthopedic consultations and knee-replacement surgery and patients' acceptance of them in the United States and in the province of Ontario, Canada.
Methods A stratified random sample of 1486 Medicare recipients (629 from the U.S. national sample, 428 from Indiana, and 429 from western Pennsylvania) and 516 people from Ontario who had been hospitalized for knee replacement between 1985 and 1989 were surveyed by mail in 1992. Patients were asked how long they had waited to see an orthopedic surgeon and to have surgery, the acceptability of these waiting times, and their overall satisfaction with surgery.
Results About 80 percent of the questionnaires were returned, but not all the respondents answered all the questions. The rate of response to specific questions was about 60 to 65 percent in both countries. The median waiting time for an initial orthopedic consultation was two weeks in the United States and four weeks in Ontario. The median waiting time for knee replacement after the operation had been planned was three weeks in the United States and eight weeks in Canada. In the United States, 95 percent of patients in the national sample considered their waiting time for surgery acceptable, as compared with 85.1 percent in Ontario. Overall satisfaction with surgery ("very or somewhat satisfied") was 85.3 percent for all U.S. respondents and 83.5 percent for Canadian respondents.
Conclusions Waiting times for initial orthopedic consultation and for knee-replacement surgery were longer in Ontario than in the United States, but overall satisfaction with surgery was similar.
Knee replacement can be an effective treatment to alleviate pain and enhance physical functioning in patients with arthritis of the knee18. An extended wait for surgery may prolong pain and difficulties in physical function. Waiting times for knee replacement are a measure of access to an important surgical procedure for a non-life-threatening condition. We compared the duration and acceptability to patients of waiting times for orthopedic consultations and knee-replacement surgery in the United States and the province of Ontario, Canada.
Methods
Administrative Data
In the United States, hospitalizations of people over 65 years of age for knee replacement from 1985 through 1989 were identified from Medicare inpatient (Medpar) data files. Patients were excluded if their ages or places of residence were unavailable, they were enrolled in a health maintenance organization at the time of the knee replacement, they were not U.S. residents, they were known to be deceased, or their hospitalization was apparently miscoded. We identified 261,823 patients who received knee replacements and who met our criteria.
In Ontario, all hospitalizations for knee replacement for fiscal years 1985 through 1989 were identified from Hospital Medical Records Institute data files. Similar exclusion and inclusion criteria were applied. In addition, the Ontario sample included only patients with a diagnosis of osteoarthritis. We identified 9039 patients who met our criteria, of whom 24.9 percent were under 65 years of age.
Sampling Frame and Determination of Sample Size
In the United States, we sampled patients from three areas: Indiana, western Pennsylvania, and the nation as a whole. Six sampling strata were defined: rural whites <80 years old, urban whites <80 years old, rural whites 
80 years old, urban whites 80 years old, blacks, and patients of other or unknown race. Race was not recorded in the Ontario data, and Hospital Medical Records Institute files included data on patients under 65 years of age who underwent knee replacement. A lower age (75 years) was therefore chosen for stratification. Four Ontario sampling strata were defined: rural patients <75 years old, urban patients <75 years old, rural patients 75 years old, and urban patients 75 years old.
In total, 1750 patients were randomly selected in the United States (750 from the national sample, 500 from the Indiana sample, and 500 from the western Pennsylvania sample). In Ontario, because of data-linkage difficulties, we were able to obtain the addresses of only 648 of the 800 randomly selected patients. To examine the effects of race, age, and residence, the sizes of the strata were set to equalize the variances of the strata means19. Sample sizes for every year of the five years of the study period were approximately equal.
Survey Procedures
All the identified patients were mailed a questionnaire in 1992. Those who did not respond to two mailings or who returned an incomplete survey were interviewed by telephone. The study was approved by the institutional review boards of the University of Toronto, Indiana University, and the Research Triangle Institute. Participants provided written informed consent.
Patients were asked about knee-replacement procedures they had undergone between 1985 and 1989. They were asked the following questions: Once you had located the surgeon, how many weeks passed before you saw the surgeon for the first time? And once you and your surgeon decided to go ahead with the knee replacement, how many weeks passed before the operation was done? Patients rated these periods of time as "too soon," "acceptable," or "too long." Patients rated their overall satisfaction with the knee replacement in five categories ranging from "very satisfied" to "very dissatisfied." If a patient had undergone multiple procedures, questions were confined to the first procedure.
Statistical Analysis
Results were adjusted for the sampling design with weighting based on the size of each stratum and the response rates associated with it19. Wilcoxon rank-sum tests, t-tests, and Pearson correlation coefficients were used for bivariate comparisons, with two-tailed P values20. Weighted linear regression and multiple logistic-regression analyses21,22 were used to assess the effect of the following independent variables on surgical waiting times and their acceptability to patients4: the age, sex, income, education, and race of the patients; the volume of knee replacements, teaching status, and number of beds of the hospitals; clinical factors (an index of coexisting conditions generated by Patient Management Category software [Pittsburgh Research Institute],23 the type of arthritis, and the condition of the knee before surgery, as reported by the patient); and geographic and temporal factors (the year of surgery and whether the patient was from an urban or rural area). The condition of the knee before surgery was assessed according to the degree of knee pain at rest (sitting or lying down) and ability to climb stairs. Since waits were skewed toward shorter waits, their logarithm (base 10) was used as the dependent variable, with the minimal wait set at one week. We performed statistical analyses using SAS (version 6.04, SAS Institute) and Sudaan (version 6.34, Research Triangle Institute) software.
Validity of the Survey
To assess the validity of a patient's recall of events surrounding a past knee-replacement operation,24 a random sample of 126 patients who had had such surgery between 1984 and 1990 were surveyed and their medical records abstracted. The level of agreement between the survey responses and the medical records, according to the criteria of Landis and Koch,25 was "poor" to "fair" for recall of pain and function (weighted kappa, 0.19 to 0.32) and "moderate" to "almost perfect" for recall of events before knee replacement, such as prior surgery, use of medications, occupational status, and living circumstances (weighted kappa, 0.41 to 0.98). In the pilot study, no significant relation was found between the level of agreement and sex, age, or time since knee replacement. Information on waiting times was not included in the medical records.
Results
After 232 deceased and 32 ineligible patients had been excluded, 1486 patients were surveyed in the United States (national survey, 629; Indiana, 428; and western Pennsylvania, 429). After 90 deceased patients, 42 ineligible patients, and 152 patients for whom we could not obtain addresses had been excluded, 516 patients were surveyed in Ontario. We obtained 1193 survey responses in the United States, and 430 in Ontario. Not all the respondents answered all the questions; therefore, the actual response rate for specific questions, as a percentage of the mailed questionnaire, was about 60 to 65 percent in both countries. As compared with those who did not respond, respondents were more likely to be younger (74.8 vs. 75.8 years), to be white (72.1 percent vs. 62.1 percent), to have been sent home under their own care (74.9 percent vs. 65.9 percent), and to have had shorter, less costly hospitalizations (12 days and $13,310 vs. 13.3 days and $14,229).
Duration and Acceptability of Waiting Times
The median waiting time for an initial consultation was two weeks in the United States and four weeks in Ontario (Table 1). The mean waiting time was consistently shorter in each U.S. survey area than in Ontario (P = 0.009 for the national sample, and P<0.001 for Indiana and western Pennsylvania).
|
|
In both the United States and Canada, the proportion of patients who considered their waiting times to be acceptable declined as the duration of the wait increased (Figure 1 and Figure 2). For patients facing similar waiting periods, the acceptability of waiting times did not differ significantly between the countries. For example, the acceptability of waiting times for patients in the U.S. national and Ontario samples who waited no more than four weeks for knee replacement was 97.0 percent and 92.7 percent, respectively (P = 0.09). Similar results were obtained for the Indiana and western Pennsylvania samples (data not shown).
|
|
Separate regression models were used to analyze the determinants of waiting times for knee replacement in each U.S. area and Ontario; estimates are presented only for the U.S. national and Ontario samples. Factors considered in the regression models accounted for 20.5 percent of the variation in waiting times for surgery in both these samples.
Institutional factors, including the number of knee replacements performed at a hospital, whether the hospital was a teaching institution, and the number of beds, were important in explaining waiting times for surgery in Ontario, but they were insignificant in the United States. The mean waiting time for knee replacement in teaching hospitals in Ontario was 16.3 weeks, as compared with 10.4 weeks for nonteaching hospitals (P = 0.009). In Ontario 44.9 percent of knee-replacement procedures were undertaken in teaching hospitals, as compared with 16.4 percent in the U.S. national sample. The percentages of teaching-hospital beds were similar (24.1 percent in Ontario and 21.5 percent in the United States)26,27.
Income, education, and sex were not associated with waiting times for surgery in either country (data not shown). Race was not associated with waiting times in the United States (data not shown). In the United States, waiting times for people 80 years old or older were shorter than those for people younger than 80 (data not shown). The condition of the knee before surgery, as assessed by patient-reported knee pain at rest (sitting or lying down) and the ability to climb stairs, was not associated with waiting time in Ontario. In the U.S. national sample, however, people who said they were unable to climb stairs before knee-replacement surgery reported shorter average waiting times (2.4 weeks) than those who could climb stairs (5.0 weeks) (P = 0.04). Coexisting conditions and the type of arthritis were not associated with waiting times for surgery (data not shown).
No temporal trend toward longer or shorter waiting times was found in the United States or Ontario for the five-year period. Urban and rural patients in Ontario had similar waits. In the U.S. national sample, patients from urban areas reported average waits that were 66 percent longer than those of patients from rural areas (5.8 vs. 3.5 weeks, P = 0.02).
Determinants of Patients' Acceptance of Waiting Times
Separate regression models were used to analyze the determinants of patients' acceptance of waiting times in the U.S. national and Ontario samples. These models accounted for 7.6 percent of the variation in the acceptability of waiting times in the United States and 25.2 percent of the variation in Ontario.
The duration of the waiting time was the most important factor in accounting for the acceptability of waits in Ontario (P<0.001). Other factors associated with the view that the waiting time for surgery was unacceptable were the inability to use stairs (P = 0.04), older age (P = 0.002), and dissatisfaction with the overall surgical outcome (P = 0.001).
Discussion
We found that waiting times for an initial orthopedic consultation and for knee-replacement surgery, as measured by patient reports, were longer in Ontario than in the United States, but the differences were not as large as some might have anticipated. A substantial majority of respondents in both countries considered their waits acceptable. There was, however, somewhat less acceptance of these waits in Ontario than in the United States. Although we did not directly evaluate the influence of waiting times on the outcome of knee-replacement surgery or the subsequent quality of life, longer waiting times did not reduce overall satisfaction with knee replacement, as reported on the questionnaire. Patients' acceptance of long waiting times might in part reflect voluntary delays,14 since people may postpone surgery for personal reasons or may wait to be operated on by a surgeon of their choice.
Patients who reported more preoperative knee pain and limits to physical functioning were found to be less accepting of their waits, but these patient factors were not related to waiting times. If surgeons were to classify patients according to the preoperative condition of the knee and to give priority for operations to patients with knee pain and substantial functional limitations, the overall acceptability of waiting times might increase.
The main limitation of our study concerns potential recall biases. People, many of them 75 years of age or older, were asked to recall waiting times for an operation that had occurred two to seven years earlier. We surveyed a random sample of patients and evaluated waiting times over a period of several years. Recall of events surrounding knee replacement was assessed in a pilot study that included a comparison with medical records24. We were unable to find a systematic relation between the time that had elapsed since the operation and the accuracy of a patient's memory of the duration of the wait for surgery. Although the reported waiting time for any patient might be imprecise, we had no reason to suspect that those surveyed were likely to underestimate or overestimate waiting times in a systematic fashion. Finally, we used regression analyses to control for patient satisfaction and thus to minimize problems associated with recall biases attributable to patient satisfaction.
Supported by the U.S. Agency for Health Care Policy and Research under a grant (06432) to Indiana University and subgrantees. Dr. Wright holds a Medical Research Council Scholarship, and Dr. Hawker a Clinical Research Fellowship from the Canadian Arthritis Society.
We are indebted to our advisory committee for insightful comments. This article is dedicated to the memory of Geoff Coyte.
Source Information
From the Department of Health Administration and Institute for Policy Analysis, University of Toronto, the Institute for Clinical Evaluative Sciences, Sunnybrook Health Sciences Centre, and the Hearing Health Care Research Unit, all in Toronto (P.C.C.); the Department of Surgery, Division of Orthopaedics and Clinical Epidemiology, Hospital for Sick Children, Toronto (J.G.W.); the Department of Medicine, Division of Rheumatology, Women's College Hospital, Toronto (G.A.H.); the Department of Medicine, Division of Rheumatology, Wellesley Hospital, Toronto (C.B.); the Department of Medicine, Division of General Internal Medicine, Regenstrief Institute for Health Care, and Roudebush Veterans Affairs Medical Center, Indianapolis (R.S.D.); Bowen Research Center, Indiana University, Indianapolis (R.S.D., D.A.F.); the Research Triangle Institute, Research Triangle Park, N.C. (J.E.P.); and the University of Toronto, Toronto (E.H.).
Address reprint requests to Dr. Coyte at the Dept. of Health Administration, 2nd Fl., McMurrich Bldg., Faculty of Medicine, University of Toronto, 12 Queen's Park Cres. W., Toronto, ON M5S 1A8, Canada.
References
| |||||||||||||||||||||||||||||||||||||||
Related Letters:
Knee-Replacement Surgery in the United States and Ontario
Russell A.S., Sullivan D. J., Coyte P. C., Wright J. G.
Extract |
Full Text
N Engl J Med 1995;
332:822-823, Mar 23, 1995.
Correspondence
This article has been cited by other articles:
HOME | SUBSCRIBE | SEARCH | CURRENT ISSUE | PAST ISSUES | COLLECTIONS | PRIVACY | TERMS OF USE | HELP | beta.nejm.org Comments and questions? Please contact us. The New England Journal of Medicine is owned, published, and copyrighted © 2009 Massachusetts Medical Society. All rights reserved. |
Previous
