VOLUME 59 #{149}JUNE 1977 #{149}NUMBER 6
PEDIATRICS
Vol. 59 No. 6 June
1977
821
ARTICLES
Pediatrics
Changes
in Tonsillectomy
Rates
Associated
With
Feedback
and
Review
John E. Wennberg, M.D., Lewis Blowers, M.D., Robert Parker, M.D.,
and Alan M. Gittelsohn, Ph. D.
From Harvard School of Public Health, Boston; Copley Hospital, Morriscille, Vermont; and The Johns Hopkins School of Hygiene, Baltimore
ABSTRACT. Among 13 Vermont Hospital Service Areas,
tonsillectomy rates decreased over a five-year period. In 1969, the rates in seven areas exceeded the estimated United States national rate; by 1973, the average rate for all areas had declined 46% and only one area remained above the U.S. rate. Much of the change occurred after feedback of data to the Vermont State Medical Society demonstrating 1969 variations. In 12 of the 13 areas, the relationship between feedback and change in clinical practices could not be documented; however, physicians in the area with the highest rate reviewed the indications for tonsillectomy and adopted a second opinion procedure for reviewing candi-dates for the surgery. The experience suggests that feedback of population-based data on incidence of procedures may be a valuable tool for the peer review process. Pediatrics,
59:821-826, 1977, TONSILLECTOMY, POPULATION DATA, PEER
REVIEW.
receives the majority of their care from one or two local hospitals. In 1969, there was a 13-fold difference between the lowest and highest age-adjusted per capita rate of tonsillectomy among the 13 largest Hospital Service Areas. In the area with the highest rate, the 1969 rate implies tonsillectomies in 63% of the resident population by age 25.’ Information on inter-hospital service area differences in incidence rates became avail-able in February 1971, and was given to the Utilization Review Committee of the Vermont State Medical Society which includes a member from the physician staff of most of Vermont’s 18 hospitals.
The incidence of tonsillectomy has been moni-tored in all Vermont and neighboring out-of-state short-term community hospitals since 1969. For purposes of analysis of medical care practices, the state has been subdivided into geographically separate Hospital Service Areas, each of which
(Received June 22; revision accepted for publication September 22, 1976.)
Supported in part by grants from the Robert Wood Johnson
Foundation and the Commonwealth Fund through the
Center for the Analysis of Health Practices, Harvard School of Public Health, and by U.S. Public Health Service contract HRA 230-75-0192.
ADDRESS FOR REPRINTS: (JEW.) Harvard School of
Public Health, 677 Huntington Avenue, 4th Floor, Boston, Massachusetts 02115.
at Viet Nam:AAP Sponsored on September 8, 2020 www.aappublications.org/news
TOTAL NUMBER OF TONSILLECTOMIES AND ADENOIDECTOMIES ALONE AND PERCENT OF
ToNsILiCToMIEs PERFORMED LOCALLY ON CHILDREN 14 YEARS OF AGE AND YOUNGER IN 13
VERMONT HOSPITAL SERVICE AREAS (1969-1973)
Area No. of Children
No. of Procedures
A % of Tonsillectomies Performed at Local Hospital(s) (1969-1973)
,-1969 ,__.*____\ T A 1970 T A 1971 T A 1972 ,__A.__ T A-,
1973 T A1 3,424 152 2 121 2 111 5 49 4 17 15 95
2 3,264 78 1 58 5 36 4 22 3 31 9 80
3 6,828 125 2 107 3 62 2 39 11 54 25 84
4 2,255 35 2 44 1 41 1 16 5 14 3 86
5 15,583 213 6 259 5 269 6 196 3 124 28 94
6 4,417 60 11 65 7 51 15 18 13 18 11 90
7 4,486 61 1 88 2 104 3 75 2 42 2 91
8 14,570 188 17 200 11 140 25 148 .34 178 34 86
9 6,160 74 1 70 5 45 3 49 5 36 12 88
10 34,481 339 49 298 49 202 47 174 107 168 100 93
1 1 10,295 99 6 1 10 10 121 20 100 22 85 1 1 80
12 5,439 38 1 24 3 22 3 20 5 22 14 53
13 2,999 10 2 17 4 10 3 20 5 8 2 72
TABLE I
The purposes of this report are: (1) to describe trends in incidence of tonsillectomy, with or without adenoidectomy, and of adenoidectomy alone in the years 1969 through 1973 in Vermont and the United States, and variations and trends among individual Vermont Hospital Service Areas; (2) to suggest that the reporting of data on incidence to practicing physicians can stimulate and facilitate changes in medical practices; and (3) to propose that this hypothesis be tested in other areas within the context of Professional Standard Review Organizations.
METHODS
Measurement of the per capita use of surgical procedures is based on the willingness of all hospitals in Vermont and in neighboring out-of-state areas to provide a uniform discharge abstract for each patient to the Cooperative Health Information Center of Vermont. Abstracts for the years 1969 to 1973 are available for this study. They contain information on the diagnosis, operative procedure, a code for the attending physician and surgeon, and the patient’s age, sex, and town of residence. Vermont is organized administratively into 250 towns that average about 36 miles in area. Geographic areas for study were defined by assigning each Vermont town to a unique Hospital Service Area. A simple proce-dure was followed: patient records were classified
initially by town and hospital; towns were then assigned to the hospital used by 60% or more of patients. Towns that did not meet this criterion or were assigned to out-of-state hospitals are not included in this study. To avoid possible confusion between mailing address and town of residence, zip code-town border relationships were investi-gated to make certain no town was assigned to a Hospital Service Area that did not contain its post office. Thirteen areas with populations of 8,000 persons or more were defined. Tonsillectomy and adenoidectomy rates in this paper are for children under 15 years of age; our population estimate is the 1970 census.
Our estimate for the Vermont state rate is based on the number of children and procedures performed in the 13 areas. This excludes about 10% of Vermont children who reside in Hospital Service Areas with total populations less than
4,000,
or who reside in a Hospital Service Area served by an out-of-state hospital. National mci-dence rates are provided by the National Center for Health Statistics and are based on a national sample of hospital discharges. Rates are not available for 1969 and 1970 and we have esti-mated rates for these years by linear interpola-tion. For 1968, the U.S. rates for tonsillectomy with or without adenoidectomy are 141 proce-dures per 10,000 children, 14 years of age and younger. The corresponding rate forat Viet Nam:AAP Sponsored on September 8, 2020 www.aappublications.org/news
TABLE II
ARTICLES
823
ANNUAL INCIDENCE OF TONSILLECTOMY AND ADENOIDECTOMY ALONE IN VERMONT AND THE
UNITED STATES (RATES PER 10,000 CHILDREN, 14 YEARS OF AGE AND LESS, 1969-1973)
Tonsillectomy Adenoidectomy
Tonsillectomy and All Adenoidectomies
Vt. U.S. U.S./VL Vt. U.S. U.S/Vt. Vt. U.S. U.S./Vt.
1969 129 136 1.05 9 7 .77 138 143 1.02
1970 128 132 1.03 10 8 .81 138 140 1.01
1971 106 128 1.20 12 9 .72 118 137 1.16
1972 81 120 1.48 20 10 .50 100 130 1.30
1973 70 117 1.67 24 11 .47 94 128 1.31
Percent
change 1969-1973
-46% -14% . .. + 166% + 45% .. . -32% -13% ...
tomy alone is 6.2 Rates for subsequent years are presented in the Results section.3
Statistical differences among areas in cross-sectional rates are tested by the
x2
distribution employing 1 degree of freedom with the null hypothesis stating that an individual area does not differ from the average rate of all areas. The extent to which the areas vary is indicated by the coefficient of variation (SD #{247}mean) and by the range between highest and lowest rates. Changes over time are evaluated by linear trendx2.
RESULTS
Number of Tonsillectomies
In the 13 areas, 5,870 tonsillectomies, with or without adenoidectomies, and 830 adenoidec-tomies alone were performed on resident children less than 15 years of age during the five-year
period (Table I). Although the number of
procedures performed on residents of the areas include those obtained at out-of-area hospitals, in all areas most of the tonsillectomies were done at the local hospital.
Annual Trends in Rates: Vermont
Compared to the U.S.
In
1969 and 1970, tonsillectomy rates in Vermont remained constant. Beginning in 1971, the incidence rate declined and by 1973 the rate was 46% lower than in 1969 (Table II). During this period the rate of decline in Vermont was considerably greater than the national trend. Our estimates of the national rates for 1969 and 1970 are 5% and 3% higher than the corresponding Vermont observations. By 1973, the U.S. rate was 67% higher than the Vermont rate.By contrast, the rate of adenoidectomy alone increased in Vermont as well as in the U.S. The
1973 Vermont rate was 2.6 times greater than the 1969 rate. By our estimate, the corresponding increase in the U.S. rate is 1.6. By 1973, on a per capita basis, Vermont children received twice as many adenoidectomies without tonsillectomies as children throughout the United States.
Although trends of use are in opposite direc-tions, tonsillectomies and adenoidectomies are not substituted on a one-for-one basis. In Vermont, the combined rate for these two procedures in 1973 is 32% lower than the 1969 rate. Nationally, the corresponding rate of decline is 13%.
Variations and Trends
The average rate of tonsillectomy in Vermont is an average of utilization occurring in its various subareas. When the population of the state is subdivided into geographic areas defined by patterns of use of hospitals, large intrastate varia-tions in resident use of tonsillectomy become apparent (Table III). In 1969, 6 of the 13 areas display rates that are different from the state average at the 0.001 level of significance. More than a 13-fold difference distinguishes the rates in the highest area from those in the lowest area and the coefficient of variation is 0.67. By 1973, the variation (although still large) has diminished: the range of variation is 4.5, two areas are “statistical outliers” at the 0.001 level of significance, and the coefficient of variation is 0.40.
The narrowing of the differences among areas relates particularly to the trend toward lower rates among areas in excess of the state average in 1969 and to the relative stability of rates in the remaining areas. Altogether, eight areas show declines that are significant at the 0.001 level (by linear trend component of the
x
statistic). Six ofat Viet Nam:AAP Sponsored on September 8, 2020 www.aappublications.org/news
ANNUAL RATE OF TONSILLECTOMY IN 13 VERMONT HOSPITAL SERVICE Ait.s (RATES PER
10,000 CHILDREN, 14 YEARS OF AGE AND LESS, 1969-1973)
Linear Trend
Area 1969 1970 1971 1972 1973 Probability
1 444#{176} 353#{176} 324#{176} 143#{176} 50 <.001
2 239#{176} 178t 110 67 95 <.001
3 183#{176} 157 91 57 79 <.001
4 155 195t 182#{176} 71 62 <.001
5 137 166#{176} 173#{176} 126#{176} 80 <.001
6 136 147 116 41t 41 <.001
7 136 196#{176} 232#{176} 167#{176} 94 .063
8 129 137 96 102t 122#{176} .083
9 120 114 73 80 58 <.001
10 98#{176} 86#{176} 59#{176} 51#{176} 49#{176} <.001
11 96t 107 118 97 83 .239
12 70#{176} 44#{176} 40#{176} 37#{176} 40t .023
13 33#{176} 57#{176} 330 67 27t .930
Coefficient of variation
Ratio high/low No. of outliers
(P < .001)
1969
Summary Statistics
1970 1971 1972 1973
0.67 0.53 0.65 0.47 0.40
13.5 8.0 9.8 4.5 4.5
6 6 7 5 2
OP .()01.
tP .01.
No. of areas with rate 7 8 4 3 1
U.S. average
TABLE III
these areas are above the state average in 1969. Rates in Area 8, which in 1969 is at the state average, do not appear to change and by 1973 this area ranks highest in the state. Area 6 shows an increase followed by a decline after 1970.
In 1969 and 1970, seven and eight Vermont areas, respectively, had rates that we estimate to be in excess of the national average. By 1973 only one area (Area 8) is above the average.
The experience in Area 1 is of particular interest. In this area, the number of tonsillec-tomies declined 89% over the five-year period. Table IV shows by location of surgeon the number of procedures performed in each study year. The number performed out-of-area is small and does not contribute to the trend. Between 1969 and 1970 a 23% drop occurred in the number of locally performed tonsillectomies. The departure from the area in the summer of 1970 of a surgeon who undertook 26 procedures in 1969 and 10 in 1970 may account for part of this drop. In subsequent years local physicians performing tonsillectomies did not change residence or retire
from practice. Between 1971 and 1973 the number of tonsillectomies they undertook declined about 85%. Whereas the number of adenoidectomies alone increased substantially in the last study year, they are not substituted on a one-for-one basis for tonsillectomy, and the net number of procedures for removal of either or both the tonsils and the adenoids declined 73% between 1971 and 1973.
Compared to national rates, rates in Area 1 are 3.3 and 2.5 times greater for tonsillectomies in 1969 and 1971, respectively. By 1973, tonsillec-tomy rates in Area 1 were 57% lower than the national average; the combined rate of tonsillec-tomies and adenoidectomies alone is 27% lower than the national average.
DISCUSSION
To our knowledge, this is the third report of changing surgical practices associated with feed-back on incidence rates in local communities. In
1959, Lembcke4 reported changes in incidence of pelvic surgery following initiation of a peer
at Viet Nam:AAP Sponsored on September 8, 2020 www.aappublications.org/news
TABLE IV
NUMBER OF TONSILLECTOMIES AND ADENOIDECTOMIES
ALONE IN VERMONT HOSPITAL SERVICE AREA 1 BY
LOCATION OF PHYSICIANS (PERFORMED ON CHILDREN 14
YEARS OR LESS, 1969-1973)
ARTICLES
825
review process in which feedback on population incidence played a role. A committee of the Saskatchewan College of Physicians and Sur-geons5 (formed to investigate trends in per capita use of hysterectomies) developed criteria for “jus-tified” and “unjustified” hysterectomies and clas-sified hysterectomies in selected hospitals by these criteria. The committee reported a 50% reduction in hysterectomy rate in one Saskatch-ewan city and an overall provincial reduction of 33% which they attributed partly to the review.
In this report, variations among areas become less and a national trend toward decreased use of tonsillectomy appears to have accelerated follow-ing feedback, suggesting that information com-paring performances among neighboring areas may per se enhance diffusion of state-of-the-art opinion and promote greater uniformity in din-ical decision-making. But we are uncertain about this possibility. A proposed Experimental Medical Care Review program to inquire into the reasons for variation among areas was submitted in May 1971 to the Department of Health, Education, and Welfare by the Vermont State Medical Society. But funding was not available and plans to document the population-based data feedback could not be implemented. Other factors such as changes in professional staffing patterns, or educational programs unrelated to feedback may account for the accelerated decline. Similarly, we are unable to account for the lack of change in Area 8.
We have, however, documented the relation-ship of changes to feedback in Area 1. The changes prior to 1970 may relate in part to departure of an active surgeon. After 1970, the changes were indeed influenced by data feedback which led to an active review process. Two of us
(
L.B., a surgeon, and R.P., a pediatrician) are members of the staff of the hospital in Area 1. Awareness of the differences among the areas led us to review the literature on the indications for tonsillectomy, and we subsequently accepted Haggerty’s6 viewpoint on the indications for tonsillectomy and adenoidectomy. We also agreed between us to review jointly each candi-date for tonsillectomy, whether seen on referral or in our own practices. By the end of 1972 we reviewed most of the tonsillectomies performed at our local hospital. We believe this process of obtaining a second opinion helped us standardize the decision process.The hypothesis that feedback of data on mci-dence of procedures can lead to a reduction in variations and more rapid diffusion of state-of-the-art opinion needs further testing. Variations
Physician Location 1969 1970 1971 1972 1973
Tonsillectomies
Outofarea 3 6 7 6 3
Local 149 115 104 43 14
All physicians 152 121 111 49 17
Adenoidectomies
Outofarea 1 1 4 3 1
Local 1 1 1 1 14
All physicians 2 2 5 4 15
Both
Outofarea 4 7 11 9 4
Local 150 116 105 44 28
All physicians 154 123 116 53 32
in the admission rate rather than in length of stay or cost per case appear to account for the majority of variation in per capita patient days or expenditures for many common surgical and medical conditions.7 This underscores the impor-tance of developing successful methodologies to deal with variations in incidence. The opportuni-ties for development of methods depend on the availability of data and the acceptance of respon-sibility for feedback and review. The likely candi-date for accepting responsibility would appear to be the Profssional Standards Review Organiza-tions (PSRO) which have, explicitly, the responsi-bility for region-wide quality assurance. The methods of data feedback developed by PSRO should be documented and its impact on varia-tions in incidence closely monitored.
It must be noted that change alone does not indicate more rational use of medical or surgical technology. In many cases, adequate studies of end-result implications of use of alternative treat-ments have not been undertaken and it is impos-sible to fully evaluate the health status implica-tions of different levels of use of a particular procedure. In the case of tonsillectomy we are confident (but not certain) that lower rates are not associated with unattended and treatable morbidity. At least for a decade there has been a decline in the popularity of tonsillectomies in the United States, and we think the process of data feedback and review of indications has increased the diffusion of state-of-the-art information. However, Paradise8 and, subsequently, the report of the Workshop on Tonsillectomy and Adenoi-dectomy9 emphasize the need for prospective clinical trials to obtain definitive data on the
at Viet Nam:AAP Sponsored on September 8, 2020 www.aappublications.org/news
outcome effects of adenoidectomy or adenotonsil-lectomy on patients with tonsillitis, pharynoton-sillitis, and otitis media. We concur with this recommendation.
REFERENCES
1. Wennberg JE, Gittelsohn A: Small area variations in health care delivery. Science 182: 1 102, 1973. 2. Vital and Health Statistics: Surgical Operations in
Short-Stay Hospitals, United States, 1968, publication (HMS) 73-1762, series 13, No. 11. Dept of Health, Education and Welfare, 1968.
3. Monthly Vital Statistics Reports: Hospital Discharge Survey Data, publication (HRA) 75-1120, vol 24,
No. 3, supplement. Dept of Health, Education and Welfare, May 30, 1975.
4. Lembcke PA: A scientific method for medical auditing. Hospitals 33:65, 1959.
5. Dyck F: An audit of hysterectomies in Saskatchewan.
Presented at the annual meeting of the Royal College of Physicians and Surgeons of Canada, Winnipeg, Manitoba, January 1975.
6. Haggerty RJ: Diagnosis and treatment: Tonsils and
adenoids-a problem revisited. Pediatrics 41:815, 1968.
7. Wennberg JE, Gittelsohn A, Shapiro N: Health care
delivery in Maine: III. Evaluating the level of
hospital performance. J Maine Med Assoc
66(11):298, 1975.
8. Paradise JL: Why T & A remains moot. Pediatrics
49:648, 1972.
9. Workshop on tonsillectomy. Ann Otol Rhinol Laryngol
84(suppl 19):78, 1975.
ACKNOWLEDGMENT
We wish to acknowledge the assistance of the people at
The Cooperative Health Information Center of Vermont
who provided the data, especially Mr. John Senning, Ms.
Patricia Hickcox, and Ms. Karen Provost.
CHANGING
It is a paradox that health professionals, in their efforts to improve people’s health-related practices, seem to expect more of the ordinary consumer than they do of themselves. Almost all patient and consumer health education assumes, explicitly, that if people know what is most healthful, they will do it.
Perhaps the most obvious test of this assumption is to look at health professionals themselves. If knowing what is health-generating were directly related to doing, then surely we in the health field would be among the most robust in the nation, slim, agile, nonsmoking, temperate eaters of complemen-tary protein, low fat and cholesterol, low-sucrose, and nonrefined carbohydrate foods, avoiders of drugging levels of alcohol and other artificial mood-changers, evenly paced in our daily patterns. This picture is obviously
non-existent. Nor do we expect it to exist. Most will recognize that it is not much more likely for a physician earning $85,000 a year to change his life pattern than for a $6,000-a-year hospital aide to do so. However, the potential for life-style change, the array of options available to these two individuals, may differ considerably.
N. MILI0
American Journal of Public Health 66:435, 1976
Noted by Student
at Viet Nam:AAP Sponsored on September 8, 2020 www.aappublications.org/news
1977;59;821
Pediatrics
John E. Wennberg, Lewis Blowers, Robert Parker and Alan M. Gittelsohn
Changes in Tonsillectomy Rates Associated With Feedback and Review
Services
Updated Information &
http://pediatrics.aappublications.org/content/59/6/821
including high resolution figures, can be found at:
Permissions & Licensing
http://www.aappublications.org/site/misc/Permissions.xhtml
entirety can be found online at:
Information about reproducing this article in parts (figures, tables) or in its
Reprints
http://www.aappublications.org/site/misc/reprints.xhtml
Information about ordering reprints can be found online:
at Viet Nam:AAP Sponsored on September 8, 2020 www.aappublications.org/news
1977;59;821
Pediatrics
John E. Wennberg, Lewis Blowers, Robert Parker and Alan M. Gittelsohn
Changes in Tonsillectomy Rates Associated With Feedback and Review
http://pediatrics.aappublications.org/content/59/6/821
the World Wide Web at:
The online version of this article, along with updated information and services, is located on
American Academy of Pediatrics. All rights reserved. Print ISSN: 1073-0397.
American Academy of Pediatrics, 345 Park Avenue, Itasca, Illinois, 60143. Copyright © 1977 by the
been published continuously since 1948. Pediatrics is owned, published, and trademarked by the
Pediatrics is the official journal of the American Academy of Pediatrics. A monthly publication, it has
at Viet Nam:AAP Sponsored on September 8, 2020 www.aappublications.org/news