lut J Gynaecol Obstei 16: 144 149, 1978
First Trimester Abortion by Vacuum Aspiration: Interphysician Variability Eva R. Miller 1 , J. L. Wood 1 , Lidija Andolsek 2 and Majda Ogrinc-Oven 2 International Fertility Research Prosram, Research Triangle Park, North Carolina, l'SA ' Family Planning Institute, University of Ljubljana. Ljubljana, Yugoslavia
ABSTRACT Miller ER, Wood J L, Andolsek L, Ogrinc-Oven M (International Fertility Research Program, Research Triangle Park, NC, USA, and Family Planning Institute, University of Ljubljana, Ljubljana, Yugoslavia). First trimester abortion by vacuum aspiration: interphysician variability, hit J Gynaecol Obstet 16: 144-149, 1978 This study compares the performances offour physicians using two types of plastic cannulae (flexible and rigid) for 1100 vacuum aspiration procedures. The criteria for assessing physician performance by cannula type were: (a) frequency of cannula obstructions, (b) amount of retained tissue obtained by sharp curettage after vacuum aspiration, (c) estimated blood loss during the procedure and (d) time required to perform the procedure. Data for each physician were compared and yielded significantly different results. The length of time required to use the cannula differed significantly among physicians; two of them had significantly shorter cannula times with the flexible instrument. The others had the lowest distributions of blood loss and the most difference between cannula usage. Consistent grouping over several variables suggests that differences in these criteria are more likely to be attributable to subtle distinctions in operator technique than to variations in equipment.
INTRODUCTION Many carefully controlled and monitored studies have been undertaken to evaluate the safety and efficacy of induced abortion (2-4). These investigations usually compare equipment or components of the vacuum aspirator apparatus; however, differences attributable to physician skill or subtle distinctions in operator technique are rarely mentioned. T h e operator's dexterity has only been at issue with respect to the suitability of procedures for Presented at the 104th Annual Meeting of the American Public Health Association, Miami Beach, FL, October 17 21, 1976.
Int J Gynaecol Obstet 16
paramedical personnel. This study was undertaken to analyze the variability in performing first trimester vacuum aspiration procedures among four physicians who were considered to have equivalent training, experience and skill.
MATERIALS A N D M E T H O D S From December 1974 to October 1975, a comparative study of physician performance in using flexible and rigid plastic cannulae for abortion procedures was conducted. T h e patients were 1100 healthy women with pregnancies of between six and 12 weeks of gestation. O n e of these patients is omitted from this analysis because her physician could not be identified. A complete protocol detailing definitions, procedures and criteria was prepared at the International Fertility Research Program (IFRP), from whose staff a physician consultant visited the clinic and observed procedures for several days. Data on the clinical management of the abortion, as well as on the patients' medical histories and selected sociodemographic variables, were collected on standard forms. (Follow-up data were obtained from 98.5% of the patients between 14 and 90 days after the abortion.) Both the flexible and rigid plastic cannulae have rounded tips. The flexible instrument has two sharp, lateral, triangular openings on opposite sides of the distal end, with the portion overhanging each aperture acting as a curette. T h e rigid plastic cannula is beveled at the distal end (Fig. 1). Study cannulae were randomly assigned, each type being used for 550 cases. One of the four physicians performed each procedure in accordance with the routine hospital rotation. No obvious differences in training or clinical experience among the physicians were known. The degree of randomness among the subsets of patients treated by each physician was tested by comparison of selected socio-
Physician variance in vacuum aspiration
demographic characteristics (Table I). T o control for evaluator bias regarding cannula type, another physician, who was unaware of the type of cannula used in a particular case, assumed responsibility for
Fig.1. Rigid plastic cannula (top) and flexible plastic cannula (bottom).
145
the patient after she left the operating room. T h e evaluator determined the presence of, treated and recorded all complications before discharging the patient. T h e patient was placed in the lithotomy position and draped with clean sheets; the vagina and perineum were painted with an antiseptic solution of benzalkonium chloride. A weighted speculum was inserted and a single-toothed tenaculum placed on the anterior cervical lip. A paracervical block of 20 ml of 1% lidocaine was administered. T h e uterus was sounded and measured, and the cervix dilated with a Pratt dilator. The amount of dilatation (8.6, 10.5 or 12.4 mm) and the size of the cannula used (outer diameter of 8, 10 or 12 mm) were determined by the gestational age of the pregnancy (according to the following groupings of completed weeks: 8 weeks or less, 9-10 weeks, or 11-12 weeks). T h e study cannula was then inserted transcervically to the fundus and suction was initiated. Vacuum was created with an electric p u m p [Ljubljana VA-3: fully described in Andolsek (1)]. If the cannula was
Table I. Selected patient characteristics for 1099 patients treated by four physicians. Physi cian A (N = 278) Patient Characteristics Age (years) 40 Mean Parity 0 1-2 3-4 >5 Unknown Mean Marital status Single Married Formerly married Formal education (years) 13 Mean Gestational age (weeks) 6-7 8-10 11-12 Mean
%
No. 33 126 92 27
11.9 45.3 33.1 9.7
B (N=311)
27 164 95 25
28.7 70 168 34 5 1
%
No.
8.7 52.8 30.5 8.0
No.
75 209 24 3 0
1.4
%
16 127 90 24
28.3 25.2 60.4 12.2 1.8 0.4
D (N=253)
C (N = 257)
6.2 49.4 35.0 9.4
54 163 39 1 0
1.4
10.3 47.4 30.0 12.3
26 120 76 31 28.8
29.1 24.1 67.2 7.7 1.0 0.0
%
No.
21.0 63.4 15.2 0.4 0.0
23.3 65.6 11.1 0.0 0.0
59 166 28 0 0 1.4
1.5
84 182 12
30.2 65.5 4.3
88 209 14
28.3 67.2 4.5
69 175 13
26.8 68.1 5.1
67 178 8
26.5 70.4 3.1
42 202 34
15.1 72.7 12.2
41 230 40
13.2 74.0 12.8
37 189 31
14.4 73.5 12.1
29 197 27
11.5 77.9 10.6
9.7 67 159 52
9.9 24.1 57.2 18.7
8.9
9.7 13.5 74.9 11.6
42 233 36 8.9
60 176 21
9.9 23.3 68.5 8.2
8.6
19.8 70.0 10.2
50 177 26 8.7
IntJ Gynaecol Obstet 16
146
Miller et al
blocked, the operator removed it from the uterus to clear it of any obstruction. When the operator judged the uterus to be adequately evacuated, a # 2 sharp metal curette was used to check for retained products of conception, and any tissue obtained was weighed. If the operator believed that the uterus was not adequately evacuated after the curette check, further evacuation was performed. After the abortion was completed, the patient was transferred to a recovery room. Except for the prepackaged sterile plastic cannulae, which were discarded after use, all instruments were sterilized in an autoclave between procedures. Gestational age was calculated as the number of completed weeks from the first day of the patient's last normal menstrual period to the day of abortion. Immediate complications were defined as those occurring from the beginning of the procedure to hospital discharge, and follow-up complications were those occurring between hospital discharge and the follow-up contact. T h e recommended period of follow-up was from two to four weeks after the abortion; patients were considered lost to followu p after 90 days. Blood loss during the abortion was estimated by the physician from aspirated uterine contents; blood loss of more than 250 ml was defined as excessive. Cannula obstruction was defined operationally by the number of times it was removed from the uterine cavity; speculum time was the interval between initial insertion and final removal of the instrument. In the event of a complication, such as uterine perforation requiring laparotomy, any additional time needed to complete the abortion was added to the time the speculum was in place. C a n n u l a time was defined as the period from the initial insertion of the cannula to its final removal. T h e two types of cannulae and physician performance were compared according to the following criteria: (a) frequency of cannula obstructions, (b) amount of tissue obtained by sharp curettage after vacuum aspiration, (c) estimated blood loss during the procedure, (d) frequency of specific complications and (e) time required to perform the procedure. T h e probability of rejecting the hypothesis that the types of cannulae or that the performance of the four physicians were not different (a) with respect to these criteria of performance was 0.05. Gestational age was controlled for by analyzing patients in groups as follows: 6-7 weeks, 8-10 weeks and 11-12 weeks. Each group was compared by physician, type of cannula and study criteria. Healthy women from six to 12 weeks pregnant desiring induced abortion and consenting to participate in the study were selected as subjects. Patients
IntJ Gynaecol Obstei 16
with preexisting medical conditions (systemic diseases), those in whom abortion had been initiated (spontaneously or artificially), those for whom general anesthesia or concurrent surgery was a necessity and those for whom intrauterine device insertion was anticipated were excluded from the study.
RESULTS T h e degree to which known measurements (duration of pregnancy, cannula time, estimated blood loss during the procedure, amount of retained tissue obtained by curettage after the procedure and number of cannula insertions) could be used to distinguish among physicians and cannula types was tested by using stepwise discriminant analysis. In this approach, the variables which appear to have discrimination value are ranked according to their relative strength to do so. Table II lists the order in which they contributed to each discrimination. T o differentiate among physicians, the variables of cannula time, estimated blood loss and number of cannula insertions were the most effective, while duration of pregnancy and amount of retained tissue after vacuum aspiration were not useful. Only number of insertions and cannula time were effective in discerning distinctions between cannulae. Wilks' lambda (X), shown in Table II, is a measure of the discriminating power of each new combination of variables as they were added; the closer to zero the X, the more discrimination possible. T h e significance of À is determined using an F-test (see column 2). T h e F-ratio can be thought of as the ratio of the distance between groups over group cohesiveness. T h e F-test in column 4 is for each individual variable (irrespective of the others in the discriminant function). T h e ability to discriminate on physician is much stronger (X = 0.45) than the ability to discriminate on type of cannula (X = 0.97). All of the F-ratios between physicians are statistically significant at the 0.05 level, 31 091 discriminant function. All independent variables were positively correlated with one another. T h e weakest series of correlations (which were not statistically significant) were those involving retained tissue. While duration of pregnancy was observed to have a strong positive correlation with estimated blood loss, cannula time and insertions, it was not necessary to control for duration of pregnancy when comparing physicians' performances because the different distributions of duration by physician were not statistically significant (Table I). T h e rigid plastic cannula (which these physicians had routinely used before this study) tended to
Table SS . S T i t f e ^ T f c ^ ^ N ^ ^ ^ ^
. Statistic F for Individual Variable
Wilks' Independent Variable Physicians Cannula time Estimated blood loss Insertions Pregnancy duration Retained tissue All variables Cannulae Insertions Cannula time Retained tissue Estimated blood loss Pregnancy duration* All variables
X
F for Discrimination
0.60 0.51 0.45 0.45 0.45
248.3" 148.5" 144.5" 85.0" 68.3"
0.45
68.3"
0.99 0.97 0.97 0.97 -
11.0" 15.6" 10.9" 8.2" -
0.97
8.2"
df 31 62 92 122 153
091.0 188.0 660.5 889.4 012.2
248.3" 65.1" 43.8" 3.3 1.1
153 012.2
68.3"
11 21 31 41
097.0 096.0 095.0 094.0 -
10.6" 20.5" 1.3 0.1 -
41 094.0
8.2"
df 31 31 31 31 31
091.0 091.0 091.0 091.0 091.0
153 012.2 11 11 11 11
094.0 094.0 094.0 094.0 -
41 094.0
" Significant at the 0.05 level. " F was less than the stepwise criterion of 0.1, so this variable was not included in the discriminant function.
Table III. Cannulae insertions (controlled for pregnancy duration) for four physicians. One Insertion
Two Insertions
Three or More Insertions
vician
No.
%
No.
%
No.
%
Less than 8 weeks' gestation (x2 = 36.2, 3 df, significant)
A B C D
55 39 24 20
82.1 92.9 40.0 40.0
12 3 36 30
17.9 7.1 60.0 60.0
0 0 0 0
0.0 0.0 0.0 0.0
8-10 weeks' gestation (X2 = 193.7, 6 df, significant)
A B C D
112 193 57 53
70.4 82.9 32.4 49.1
38 35 58 76
23.9 15.1 32.9 6.5
9 5 61 48
5.7 2.0 34.7 44.4
More than 10 weeks' gestation (x2 = 22.5, 6 df, significant)
A B C D
19 20 3 2
36.5 55.6 14.2 7.7
18 8 6 13
34.6 22.2 28.6 50.0
15 8 12 11
28.9 22.2 57.2 42.3
Duration of Pregnancy
require fewer reinsertions than the flexible plastic cannula. This difference was statistically significant only for the lowest gestational age group. Differences among physicians by number of insertions were statistically significant both for the cannulae and in each gestational age group (Table III); physicians C and D had more insertions. T h e differences in the amounts of retained tissue obtained by curettage after the procedure were not statistically significant for patients treated by type of cannula. However, when amounts of retained tissue were compared for each physician, the differ-
ences were statistically significant for the first two gestational age groups (Table IV). Physician D had the highest incidence of retained tissue in each gestational age group. T h e mean estimated blood loss during the procedure (Table V) was statistically different among physicians, but not between cannulae. T h e two physicians (C and D) who had the longer mean cannula times and highest number of insertions had the lowest distributions of blood loss. Mean cannula times were significantly different among physicians and significantly different between types of cannuintj
Gynaecol Obstei 16
148
Miller et al
Table IV. Retained tissue (controlled for pregnancy duration) for four physicians. No Retained Tissue
Retained Tissue
Physician
No.
%
No.
%
Less than 8 weeks' gestation (X2 = 11.2,3 d/, significant)
A B C D
66 38 59 43
98.5 90.5 98.3 86.0
1 4 1 7
1.5 9.5 1.7 14.0
8-10 weeks' gestation (x2 = 38.0, 3df, significant)
A B C D
145 223 167 150
91.2 94.1 94.9 84.7
14 14 9 27
8.8 5.9 5.1 15.3
More than 10 weeks' gestation (X2 = 3.2, 3 df, not significant)
A B C D
44 33 20 21
84.6 91.7 95.2 80.8
8 3 1 5
15.4 8.3 4.8 19.2
Duration of Pregnancy
Table V. Mean estimated blood loss (in ml) by type of cannula for four physicians. Physi cian" A
Cannula
B
C
D
Flexible Mean Standard deviation
N = 127 N = 155 N=138 73.0 96.5 88.8
Nonflexible Mean Standard deviation
N = 151 N = 156 N = 119 N = 123 100.3 88.9 69.9 79.1
60.8
f Significant
63.5
49.5
N = 130 76.7 46.2
64.9
57.7
41.4
56.6
0.5 No
0.1 No
0.5 No
0.4 No
" For physicians: F = 12.1; 31 095 df, at the 0.05 level.
significant
lae for two of the physicians. It should be noted, however, that although statistically significant differences were observed, the mean range varied from 0.87 to 2.52 minutes, which may not matter in the practical situation. Differences among physicians for the interval between insertion and removal of the speculum were not statistically significant. None of the specific, immediate, follow-up or overall complication rates were statistically significant when compared by cannula type or by physician.
than between the cannulae. T h e four physicians participating in this study were experienced; their performances, measured by these criteria, were all well within acceptable ranges of safety. When statistically significant differences among these physicians were observed for any of the criteria used, it could be fairly questioned whether these differences had any practical implications. Differences among physicians in amount of tissue obtained by sharp curettage and in estimated blood loss may reflect degrees of accuracy of these estimates rather than true differences related to performance. H a d any one of these physicians been solely responsible for completing the study protocol for comparing cannulae, each would have had different results. With some equipment, testing established criteria may yield no differences; however, a physician may obtain better results using equipment he prefers. This study emphasized the importance of having different physicians perform the same protocol under controlled experimental conditions (replication) and of establishing relevant criteria to measure the efficacy of a procedure. There is an obvious need for criteria to evaluate the physician's clinical performance. There are probably subtle differences in technique which may lend themselves to measurement after more sensitive testing devices, such as one that collects and calibrates blood loss more accurately, are developed.
DISCUSSION
ACKNOWLEDGMENT
Using the criteria in this study, we found more differences among the physicians' performances
This work was supported in part by the International Fertility Research Program and the Office of
IntJ Gynaecol Obstet 16
Physician variance in vacuum aspiration
Population, United States Agency for International Development (AID/pha-C-1172).
149
3. Tietze C, Lewit S: Early medical complications of legal abortion. Joint Program for the Study of Abortion (JPSA). Stud Fam Plann 3:97, 1972. 4. V a n der Vlugt T, Piotrow P T : Uterine aspiration techniques. Population Reports, Series F, N o 3, J u n e 1973.
REFERENCES 1. A n d o l s e k L ( e d ) : T h e Ljubljana Abortion Study 1971-1973. National Institute of Health, Ljubljana, Yugoslavia, 1974. 2. Antonovski L, Sukarov L, Brenner W E , Edelman DA, Bernard R P : A comparative study of metal and plastic (Karman) cannulae for first trimester abortion by suction curettage. Int J Gynaecol Obstet 13:33, 1975.
Address for reprints: International Fertility Rearch Program Research Triangle Park, NC 27709 USA
Inl J Gynaecol Obstei 16
First Trimester Abortion by Vacuum Aspiration: Interphysician Variability Eva R. Miller 1 , J. L. Wood 1 , Lidija Andolsek 2 and Majda Ogrinc-Oven 2 International Fertility Research Prosram, Research Triangle Park, North Carolina, l'SA ' Family Planning Institute, University of Ljubljana. Ljubljana, Yugoslavia
ABSTRACT Miller ER, Wood J L, Andolsek L, Ogrinc-Oven M (International Fertility Research Program, Research Triangle Park, NC, USA, and Family Planning Institute, University of Ljubljana, Ljubljana, Yugoslavia). First trimester abortion by vacuum aspiration: interphysician variability, hit J Gynaecol Obstet 16: 144-149, 1978 This study compares the performances offour physicians using two types of plastic cannulae (flexible and rigid) for 1100 vacuum aspiration procedures. The criteria for assessing physician performance by cannula type were: (a) frequency of cannula obstructions, (b) amount of retained tissue obtained by sharp curettage after vacuum aspiration, (c) estimated blood loss during the procedure and (d) time required to perform the procedure. Data for each physician were compared and yielded significantly different results. The length of time required to use the cannula differed significantly among physicians; two of them had significantly shorter cannula times with the flexible instrument. The others had the lowest distributions of blood loss and the most difference between cannula usage. Consistent grouping over several variables suggests that differences in these criteria are more likely to be attributable to subtle distinctions in operator technique than to variations in equipment.
INTRODUCTION Many carefully controlled and monitored studies have been undertaken to evaluate the safety and efficacy of induced abortion (2-4). These investigations usually compare equipment or components of the vacuum aspirator apparatus; however, differences attributable to physician skill or subtle distinctions in operator technique are rarely mentioned. T h e operator's dexterity has only been at issue with respect to the suitability of procedures for Presented at the 104th Annual Meeting of the American Public Health Association, Miami Beach, FL, October 17 21, 1976.
Int J Gynaecol Obstet 16
paramedical personnel. This study was undertaken to analyze the variability in performing first trimester vacuum aspiration procedures among four physicians who were considered to have equivalent training, experience and skill.
MATERIALS A N D M E T H O D S From December 1974 to October 1975, a comparative study of physician performance in using flexible and rigid plastic cannulae for abortion procedures was conducted. T h e patients were 1100 healthy women with pregnancies of between six and 12 weeks of gestation. O n e of these patients is omitted from this analysis because her physician could not be identified. A complete protocol detailing definitions, procedures and criteria was prepared at the International Fertility Research Program (IFRP), from whose staff a physician consultant visited the clinic and observed procedures for several days. Data on the clinical management of the abortion, as well as on the patients' medical histories and selected sociodemographic variables, were collected on standard forms. (Follow-up data were obtained from 98.5% of the patients between 14 and 90 days after the abortion.) Both the flexible and rigid plastic cannulae have rounded tips. The flexible instrument has two sharp, lateral, triangular openings on opposite sides of the distal end, with the portion overhanging each aperture acting as a curette. T h e rigid plastic cannula is beveled at the distal end (Fig. 1). Study cannulae were randomly assigned, each type being used for 550 cases. One of the four physicians performed each procedure in accordance with the routine hospital rotation. No obvious differences in training or clinical experience among the physicians were known. The degree of randomness among the subsets of patients treated by each physician was tested by comparison of selected socio-
Physician variance in vacuum aspiration
demographic characteristics (Table I). T o control for evaluator bias regarding cannula type, another physician, who was unaware of the type of cannula used in a particular case, assumed responsibility for
Fig.1. Rigid plastic cannula (top) and flexible plastic cannula (bottom).
145
the patient after she left the operating room. T h e evaluator determined the presence of, treated and recorded all complications before discharging the patient. T h e patient was placed in the lithotomy position and draped with clean sheets; the vagina and perineum were painted with an antiseptic solution of benzalkonium chloride. A weighted speculum was inserted and a single-toothed tenaculum placed on the anterior cervical lip. A paracervical block of 20 ml of 1% lidocaine was administered. T h e uterus was sounded and measured, and the cervix dilated with a Pratt dilator. The amount of dilatation (8.6, 10.5 or 12.4 mm) and the size of the cannula used (outer diameter of 8, 10 or 12 mm) were determined by the gestational age of the pregnancy (according to the following groupings of completed weeks: 8 weeks or less, 9-10 weeks, or 11-12 weeks). T h e study cannula was then inserted transcervically to the fundus and suction was initiated. Vacuum was created with an electric p u m p [Ljubljana VA-3: fully described in Andolsek (1)]. If the cannula was
Table I. Selected patient characteristics for 1099 patients treated by four physicians. Physi cian A (N = 278) Patient Characteristics Age (years) 40 Mean Parity 0 1-2 3-4 >5 Unknown Mean Marital status Single Married Formerly married Formal education (years) 13 Mean Gestational age (weeks) 6-7 8-10 11-12 Mean
%
No. 33 126 92 27
11.9 45.3 33.1 9.7
B (N=311)
27 164 95 25
28.7 70 168 34 5 1
%
No.
8.7 52.8 30.5 8.0
No.
75 209 24 3 0
1.4
%
16 127 90 24
28.3 25.2 60.4 12.2 1.8 0.4
D (N=253)
C (N = 257)
6.2 49.4 35.0 9.4
54 163 39 1 0
1.4
10.3 47.4 30.0 12.3
26 120 76 31 28.8
29.1 24.1 67.2 7.7 1.0 0.0
%
No.
21.0 63.4 15.2 0.4 0.0
23.3 65.6 11.1 0.0 0.0
59 166 28 0 0 1.4
1.5
84 182 12
30.2 65.5 4.3
88 209 14
28.3 67.2 4.5
69 175 13
26.8 68.1 5.1
67 178 8
26.5 70.4 3.1
42 202 34
15.1 72.7 12.2
41 230 40
13.2 74.0 12.8
37 189 31
14.4 73.5 12.1
29 197 27
11.5 77.9 10.6
9.7 67 159 52
9.9 24.1 57.2 18.7
8.9
9.7 13.5 74.9 11.6
42 233 36 8.9
60 176 21
9.9 23.3 68.5 8.2
8.6
19.8 70.0 10.2
50 177 26 8.7
IntJ Gynaecol Obstet 16
146
Miller et al
blocked, the operator removed it from the uterus to clear it of any obstruction. When the operator judged the uterus to be adequately evacuated, a # 2 sharp metal curette was used to check for retained products of conception, and any tissue obtained was weighed. If the operator believed that the uterus was not adequately evacuated after the curette check, further evacuation was performed. After the abortion was completed, the patient was transferred to a recovery room. Except for the prepackaged sterile plastic cannulae, which were discarded after use, all instruments were sterilized in an autoclave between procedures. Gestational age was calculated as the number of completed weeks from the first day of the patient's last normal menstrual period to the day of abortion. Immediate complications were defined as those occurring from the beginning of the procedure to hospital discharge, and follow-up complications were those occurring between hospital discharge and the follow-up contact. T h e recommended period of follow-up was from two to four weeks after the abortion; patients were considered lost to followu p after 90 days. Blood loss during the abortion was estimated by the physician from aspirated uterine contents; blood loss of more than 250 ml was defined as excessive. Cannula obstruction was defined operationally by the number of times it was removed from the uterine cavity; speculum time was the interval between initial insertion and final removal of the instrument. In the event of a complication, such as uterine perforation requiring laparotomy, any additional time needed to complete the abortion was added to the time the speculum was in place. C a n n u l a time was defined as the period from the initial insertion of the cannula to its final removal. T h e two types of cannulae and physician performance were compared according to the following criteria: (a) frequency of cannula obstructions, (b) amount of tissue obtained by sharp curettage after vacuum aspiration, (c) estimated blood loss during the procedure, (d) frequency of specific complications and (e) time required to perform the procedure. T h e probability of rejecting the hypothesis that the types of cannulae or that the performance of the four physicians were not different (a) with respect to these criteria of performance was 0.05. Gestational age was controlled for by analyzing patients in groups as follows: 6-7 weeks, 8-10 weeks and 11-12 weeks. Each group was compared by physician, type of cannula and study criteria. Healthy women from six to 12 weeks pregnant desiring induced abortion and consenting to participate in the study were selected as subjects. Patients
IntJ Gynaecol Obstei 16
with preexisting medical conditions (systemic diseases), those in whom abortion had been initiated (spontaneously or artificially), those for whom general anesthesia or concurrent surgery was a necessity and those for whom intrauterine device insertion was anticipated were excluded from the study.
RESULTS T h e degree to which known measurements (duration of pregnancy, cannula time, estimated blood loss during the procedure, amount of retained tissue obtained by curettage after the procedure and number of cannula insertions) could be used to distinguish among physicians and cannula types was tested by using stepwise discriminant analysis. In this approach, the variables which appear to have discrimination value are ranked according to their relative strength to do so. Table II lists the order in which they contributed to each discrimination. T o differentiate among physicians, the variables of cannula time, estimated blood loss and number of cannula insertions were the most effective, while duration of pregnancy and amount of retained tissue after vacuum aspiration were not useful. Only number of insertions and cannula time were effective in discerning distinctions between cannulae. Wilks' lambda (X), shown in Table II, is a measure of the discriminating power of each new combination of variables as they were added; the closer to zero the X, the more discrimination possible. T h e significance of À is determined using an F-test (see column 2). T h e F-ratio can be thought of as the ratio of the distance between groups over group cohesiveness. T h e F-test in column 4 is for each individual variable (irrespective of the others in the discriminant function). T h e ability to discriminate on physician is much stronger (X = 0.45) than the ability to discriminate on type of cannula (X = 0.97). All of the F-ratios between physicians are statistically significant at the 0.05 level, 31 091 discriminant function. All independent variables were positively correlated with one another. T h e weakest series of correlations (which were not statistically significant) were those involving retained tissue. While duration of pregnancy was observed to have a strong positive correlation with estimated blood loss, cannula time and insertions, it was not necessary to control for duration of pregnancy when comparing physicians' performances because the different distributions of duration by physician were not statistically significant (Table I). T h e rigid plastic cannula (which these physicians had routinely used before this study) tended to
Table SS . S T i t f e ^ T f c ^ ^ N ^ ^ ^ ^
. Statistic F for Individual Variable
Wilks' Independent Variable Physicians Cannula time Estimated blood loss Insertions Pregnancy duration Retained tissue All variables Cannulae Insertions Cannula time Retained tissue Estimated blood loss Pregnancy duration* All variables
X
F for Discrimination
0.60 0.51 0.45 0.45 0.45
248.3" 148.5" 144.5" 85.0" 68.3"
0.45
68.3"
0.99 0.97 0.97 0.97 -
11.0" 15.6" 10.9" 8.2" -
0.97
8.2"
df 31 62 92 122 153
091.0 188.0 660.5 889.4 012.2
248.3" 65.1" 43.8" 3.3 1.1
153 012.2
68.3"
11 21 31 41
097.0 096.0 095.0 094.0 -
10.6" 20.5" 1.3 0.1 -
41 094.0
8.2"
df 31 31 31 31 31
091.0 091.0 091.0 091.0 091.0
153 012.2 11 11 11 11
094.0 094.0 094.0 094.0 -
41 094.0
" Significant at the 0.05 level. " F was less than the stepwise criterion of 0.1, so this variable was not included in the discriminant function.
Table III. Cannulae insertions (controlled for pregnancy duration) for four physicians. One Insertion
Two Insertions
Three or More Insertions
vician
No.
%
No.
%
No.
%
Less than 8 weeks' gestation (x2 = 36.2, 3 df, significant)
A B C D
55 39 24 20
82.1 92.9 40.0 40.0
12 3 36 30
17.9 7.1 60.0 60.0
0 0 0 0
0.0 0.0 0.0 0.0
8-10 weeks' gestation (X2 = 193.7, 6 df, significant)
A B C D
112 193 57 53
70.4 82.9 32.4 49.1
38 35 58 76
23.9 15.1 32.9 6.5
9 5 61 48
5.7 2.0 34.7 44.4
More than 10 weeks' gestation (x2 = 22.5, 6 df, significant)
A B C D
19 20 3 2
36.5 55.6 14.2 7.7
18 8 6 13
34.6 22.2 28.6 50.0
15 8 12 11
28.9 22.2 57.2 42.3
Duration of Pregnancy
require fewer reinsertions than the flexible plastic cannula. This difference was statistically significant only for the lowest gestational age group. Differences among physicians by number of insertions were statistically significant both for the cannulae and in each gestational age group (Table III); physicians C and D had more insertions. T h e differences in the amounts of retained tissue obtained by curettage after the procedure were not statistically significant for patients treated by type of cannula. However, when amounts of retained tissue were compared for each physician, the differ-
ences were statistically significant for the first two gestational age groups (Table IV). Physician D had the highest incidence of retained tissue in each gestational age group. T h e mean estimated blood loss during the procedure (Table V) was statistically different among physicians, but not between cannulae. T h e two physicians (C and D) who had the longer mean cannula times and highest number of insertions had the lowest distributions of blood loss. Mean cannula times were significantly different among physicians and significantly different between types of cannuintj
Gynaecol Obstei 16
148
Miller et al
Table IV. Retained tissue (controlled for pregnancy duration) for four physicians. No Retained Tissue
Retained Tissue
Physician
No.
%
No.
%
Less than 8 weeks' gestation (X2 = 11.2,3 d/, significant)
A B C D
66 38 59 43
98.5 90.5 98.3 86.0
1 4 1 7
1.5 9.5 1.7 14.0
8-10 weeks' gestation (x2 = 38.0, 3df, significant)
A B C D
145 223 167 150
91.2 94.1 94.9 84.7
14 14 9 27
8.8 5.9 5.1 15.3
More than 10 weeks' gestation (X2 = 3.2, 3 df, not significant)
A B C D
44 33 20 21
84.6 91.7 95.2 80.8
8 3 1 5
15.4 8.3 4.8 19.2
Duration of Pregnancy
Table V. Mean estimated blood loss (in ml) by type of cannula for four physicians. Physi cian" A
Cannula
B
C
D
Flexible Mean Standard deviation
N = 127 N = 155 N=138 73.0 96.5 88.8
Nonflexible Mean Standard deviation
N = 151 N = 156 N = 119 N = 123 100.3 88.9 69.9 79.1
60.8
f Significant
63.5
49.5
N = 130 76.7 46.2
64.9
57.7
41.4
56.6
0.5 No
0.1 No
0.5 No
0.4 No
" For physicians: F = 12.1; 31 095 df, at the 0.05 level.
significant
lae for two of the physicians. It should be noted, however, that although statistically significant differences were observed, the mean range varied from 0.87 to 2.52 minutes, which may not matter in the practical situation. Differences among physicians for the interval between insertion and removal of the speculum were not statistically significant. None of the specific, immediate, follow-up or overall complication rates were statistically significant when compared by cannula type or by physician.
than between the cannulae. T h e four physicians participating in this study were experienced; their performances, measured by these criteria, were all well within acceptable ranges of safety. When statistically significant differences among these physicians were observed for any of the criteria used, it could be fairly questioned whether these differences had any practical implications. Differences among physicians in amount of tissue obtained by sharp curettage and in estimated blood loss may reflect degrees of accuracy of these estimates rather than true differences related to performance. H a d any one of these physicians been solely responsible for completing the study protocol for comparing cannulae, each would have had different results. With some equipment, testing established criteria may yield no differences; however, a physician may obtain better results using equipment he prefers. This study emphasized the importance of having different physicians perform the same protocol under controlled experimental conditions (replication) and of establishing relevant criteria to measure the efficacy of a procedure. There is an obvious need for criteria to evaluate the physician's clinical performance. There are probably subtle differences in technique which may lend themselves to measurement after more sensitive testing devices, such as one that collects and calibrates blood loss more accurately, are developed.
DISCUSSION
ACKNOWLEDGMENT
Using the criteria in this study, we found more differences among the physicians' performances
This work was supported in part by the International Fertility Research Program and the Office of
IntJ Gynaecol Obstet 16
Physician variance in vacuum aspiration
Population, United States Agency for International Development (AID/pha-C-1172).
149
3. Tietze C, Lewit S: Early medical complications of legal abortion. Joint Program for the Study of Abortion (JPSA). Stud Fam Plann 3:97, 1972. 4. V a n der Vlugt T, Piotrow P T : Uterine aspiration techniques. Population Reports, Series F, N o 3, J u n e 1973.
REFERENCES 1. A n d o l s e k L ( e d ) : T h e Ljubljana Abortion Study 1971-1973. National Institute of Health, Ljubljana, Yugoslavia, 1974. 2. Antonovski L, Sukarov L, Brenner W E , Edelman DA, Bernard R P : A comparative study of metal and plastic (Karman) cannulae for first trimester abortion by suction curettage. Int J Gynaecol Obstet 13:33, 1975.
Address for reprints: International Fertility Rearch Program Research Triangle Park, NC 27709 USA
Inl J Gynaecol Obstei 16
