- Email: [email protected]
Cervical cancer survival in Nijmegen region, The Netherlands, 1970–1985
GYNECOLOGIC
ONCOLOGY
30, 51-56 (1988)
Cervical Cancer Survival in Nijmegen Region, The Netherlands, 1970-l 985 YOLANDA VAN DER GRAAF, M.D., PH.D. ,* PETRONELLA G.M. PEER, GERHARD A. ZIELHUIS, M.Sc., PHD.,* AND PETER G. VOOIJS M.D., Departments
of *Social Medicine and fpathology, Geert Grooteplein Nijmegen, The Netherlands
M.Sc. ,* PH.DP
Zuid 24, 6525 Ga
Received December 29, 1986 Survival rates were computed for 359 women diagnosed with invasive cervical cancer between 1970 and 1985. The 5-year survival rate for the entire group was 67%. Survival was better in the period 1976-1980. Extension of the tumor and age at diagnosis were important prognostic factors. The effects of clinical stage, age at diagnosis, and year of diagnosis were studied simultaneously with the proportional hazards model. The hazard rate increased with increasing age and increasing clinical stage. Year of diagnosis had effect on survival only in case of IIB tumors. For the other clinical stages there was no significant effect on survival of year of diagnosis. 0 1988 Academic Press, Inc.
INTRODUCTION Cervical mortality rates are declining in several Western countries. [l]. In The Netherlands cervical cancer mortality has decreased since 1962. Since cervical cancer screening programs in The Netherlands did not start before 1976, cervical screening cannot be an explanation for this decline. A fall in the incidence of cervical cancer or a shift toward diagnosis of invasive cervical cancer in a less advanced stage seems the best explanation for this “natural” decline. However improvements of therapy with a better patient survival could also account at least for some of the decrease in mortality from cervical cancer. Ketting computed survival rates for all patients with invasive cancer of the uterine cervix treated in a large municipal hospital for the calendar period 1950-1979 [2]. He reported an improvement in therapeutic results with stage I, II, and III cervical cancer. He proposed the change in the treatment scheme in the period 1961-1969 as an explanation for the decrease in mortality. In this study we evaluated the survival of cervical cancer patients in the Nijmegen region for the calendar period 19701985 in order to investigate improvement in therapeutic results in the last 15 years. PATIENTS AND METHODS Population screening for cervical cancer started in 1976 in three pilot regions in The Netherlands. In these regions a quarter of the Dutch population is living. 51 00908258188 $1.50 Copyright 0 1988 by Academic Press, Inc. All rights of reproduction in any form reserved.
52
VAN
DER
GRAAF
ET AL.
All women aged 35 through 54 were invited every 3 years for a cervical smear. Apart from the screening program an increasing number of smears are being made by general practitioners and gynecologists. In the region of Nijmegen (about 750,000inhabitants) data on the incidence of invasive cervical cancer were available from the local cancer registry which has been kept since 1970 [3]. From the files of all pathology laboratories in the region all women were registered who had a histological diagnosis of microinvasive or invasive cancer of the uterine cervix. A histological diagnosis made in a pathology laboratory outside the region was not available for registration, but such cases were exceptions. Items registered were age at diagnosis, date of diagnosis, and clinical stage of the cervical cancer. Classification of the extension of the tumor is done according to the FIG0 classification (4). All diagnoses of cervical cytological smears made in the screening program as well as by general practitioners and gynecologists were added to the registry of histological diagnoses. All patients known to have cervical cancer were followed for vital status with the help of the registrar’s offices and the Central Bureau of Statistics. For a few cases this information was obtained directly from gynecologists and general practitioners. Since several patients were alive at the time the study was terminated and other patients were lost to follow up, survival curves were computed according to the technique of Kaplan-Meier [5]. Death from other causes is considered censoring. The standard errors of the estimated survival rates were computed according to the Greenwood formula (6). To test the equality of survival curves for different groups, the Mantel log rank test was used (7). In order to assess the effect of clinical stage of disease, age at diagnosis, and year at diagnosis simultaneously we used the proportional hazards model according to cox [8]. RESULTS During the study period 378 women were diagnosed with invasive cervical cancer. No follow-up data could be traced for 10 women and for 9 women the clinical stage of the tumor remained unknown. These women were left out of the analysis. Table 1 summarizes the clinical stage distribution according to age and point of time at diagnosis for the remaining 359 women. The Kaplan-Meier survival curve for the total group of women is displayed in Fig. 1. Five-year survival was calculated at 67.4% (standard error 2.6%). This plot illustrates that j-year survival is a good parameter for survival since only a few women die from cervical cancer after the fifth year. Table 2 shows the effect of extensions of the tumor, age, and calendar period on j-year survival. In general women with less advanced stages have a better survival. Exceptions are women with cervical cancers in clinical stage IIA. These women had a better survival then women with clinical stage IB tumors, although not significantly (Mantel log rank test, P = 0.52). All other pairwise comparisons yielded significant results with the log rank test. Survival curves for each stage are plotted in Fig. 2.
CERVICAL
CANCER
SURVIVAL
IN
TABLE CLINICAL
STAGE DISTRIBUTION
THE
53
NETHERLANDS
1
ACCORDING TO AGE AND CALENDAR
PERIOD
Clinical stage IA
IB
IIA
IIB
III
IV
Total
71(35%) 13(8%)
54(26%) 28(18%)
18(9%) 12(8%)
31(15%) 51(33%)
23(11%) 42(27%)
8(4%) 8(5%)
205 154
27(19%) 38(28%) 19(25%)
35(24%) 25118%) 22(29%)
13(9%) 14(10%) 3(4%)
41(28%) 28(20%) 13(17%)
23(16%) 27(20%) 15(19%)
6(4%) 5(4%) 5(6%)
145 137 77
Age <54 255 Calendar period 1970-1975 1976-1980 1981-1985
Survival rates differed also between age groups as is shown in Table 2. Survival was better in younger women. Although survival rates differed significantly for the three calendar periods, they do not show a trend. To study the effect of clinical stage, age at diagnosis, and year of diagnosis simultaneously, these three factors were incorporated into the proportional hazards model according to Cox. This multivariate regression model takes into account the correlation among the three factors. Because it was unlikely that for the same tumor stage, improvement of therapy differed for different ages and that the effect of age on survival was different for the various clinical stages, we did not take into account interaction of age with year of diagnosis and tumor stage. It could be, however, that therapy improved in the course of time only for some clinical stages. For that reason, age at diagnosis, clinical stage, year at diagnosis, and the interactions of clinical stage and year of diagnosis were entered in the model. The effect of year of diagnosis was significant only in clinical stage IIB and so in the final model age at diagnosis, clinical stage and the effect of year at diagnosis in clinical stage IIB were entered (Table 3). Increasing age and increasing clinical stage gave a significantly higher mortality. For stage IB and IIA, however, there was no significant difference in survival. The survival rate improved over the years for women with a stage IIB tumor. DISCUSSION The overall 5-year survival rate calculated in this study population is comparable with survival rates reported by others [2,9,10]. The “extension” (clinical stage) of the tumor seems the best predictor of survival. Less advanced clinical stageswith better survival were more frequently diagnosed in young women. Only part of these better survival rates in the youngest age groups can be explained by this phenomenon since after correcting for clinical stage the effect of age remained. Survival improved significantly during the last 15 years only for women with a stage IIB tumor. In the 1980s survival rates for these women became as high as for women with IB and IIA tumors, while in the period before 1980 the 5-year survival was calculated at 46%. This gain in
VAN
survival
DER GRAAF
ET AL.
__~--
80
60
50
I 48
I 24
I 96
I 72
I 120
I 144 follow
I 168
up
I 192 (months)
FIG. 1. Survival curve of women with cervical cancer. Region of Nijmegen, The Netherlands, 1970-1985.
TABLE 2 S-YEAR SURVIVAL RATES ACCORDING TO CLINICAL STAGE, AGE GROUP, AND CALENDAR PERIOD
Clinical stage (FIGO):
Age group:
IA
IB
IIA
IIB
III
IV
97%
77%
83%
55%
36%
11%
35-54
<35 91%
Calendar period:
55-70
72%
60%
>70 34%
1970-1975
1976-1980
1981-1985
62%
72%
67%
CERVICAL
CANCER
SURVIVAL
IN
THE
55
NETHERLANDS
survival
60
40
-7
-1
I L---
C------.
m
1
20
I L------------------X
Ip
1
1
I
I
12
24
36
48
I
60
/
72 follow-up
I
84 ( months
1
FIG. 2. Survival curves of women with cervical cancer according to clinical stage. Region of Nijmegen, The Netherlands, 1970-1985.
life expectancy must be the result of improvement in radiotherapeutic treatment since all women with a stage IIB tumor were treated with radiation therapy with curative intent. This finding is comparable with the results described by Ketting [2]. In that study the 5-year survival was 40% for stage IIB patients in the period 1961-1969 and 70% in the period 1970-1974 and comparable with the 5-year survival rate in stage IIA patients. The 5-year survival rate for cases with a IB tumor was somewhat lower in our study than that reported by Ketting [2]. In his study the 5-year survival for these cases was 84% in the period 1970-1974. Possibly this difference can be explained by the fact that the study populations were not comparable. Our study was population based and the women were treated in the regional hospitals and in one university hospital, while in the study of Ketting all women were treated in a center. The population screening led to a decrease of the incidence of invasive cervical cancer. In the period prior to the population screening program (1970-1976) the yearly incidence of invasive cervical cancer was 18.6 per 100,000 in women aged 35 through 54 years. During the first screening period the incidence was 21.5 per 100,000 women, in the second screening period 9.0 per 100,000 women and
56
VAN DER GRAAF ET AL. TABLE 3
REGRESSIONCOEFFICIENTSFOR CLINICAL STAGE, AGE AT DIAGNOSIS, AND THE EFFECT OF YEAR OF DIAGNOSIS IN STAGE IIB IN Cox’s PROPORTIONALHAZARDS MODEL
Variable Age at diagnosis (reference age 50) Year of diagnosis in stage IIB (reference 1975) Clinical stage (reference IB) IA IIA IIB III IV
Standard error
Quotient
P value (one sided)
0.0078
2.34
0.0097
-0.10
0.0467
2.20
0.0137
- 2.30 -0.33 0.65 1.26 2.20
0.7421 0.4635 0.2820 0.2762 0.3716
3.10 0.70 2.32 4.59 5.92
0.0009 0.2408 0.0102 0.0000 0.0000
Regression coefficient 0.018
in the third screening period 3.3 per 100,000 women. The incidence of invasive cancer among women older than 54 years, who were not included in the cervical cancer screening program, showed a decline 6 years after the start of the program. This can be explained by the increasing percentage of women in this age group who in previous years were eligible for the cervical cancer screening program and hence received treatment when a preinvasive lesion was discovered. In the course of 15 years a shift to more favorable clinical stages became evident. This shift to less advanced stages of cervical cancer at diagnosis and the improvement of therapy for women with a stage IIB tumor are responsible for a better overall survival. For the other clinical stages survival did not improve significantly during the last 15 years. ACKNOWLEDGMENT We are grateful to Jeep Hopstaken for collecting the data.
REFERENCES 1. Hill, G. B. Cancer of the uterus mortality trends since 1950, WHO Chron. 30, 188-193 (1976). 2. Ketting, B. W. Surgical treatment of invasive carcinoma of the uterine cervix, Thesis, University of Amsterdam, Rodopi, Amsterdam (1981). 3. van der Graaf, Y., Klinkhamer, P. J. J. M., and Vooijs, G. P. Effect of population screening for cancer of the uterine cervix in Nijmegen, The Netherlands, Preven. Med. 15, 682-690 4. Annual Report on the results of treatment in carcinoma of the uterus, vagina and ovary (H. L. Kottmeier, Ed.), Vol. 15, Stockholm (1971). 5. Kaplan, E. L., and Meier P. Non parametric estimation from incomplete observations. Amer. Stat. Assoc. J. 53, 457-481 (1958). 6. Lee, E. T. Statistical methods for survival data analysis, Univ. of Oklahoma Press, Oklahoma City, 1980. 7. Mantel N. Evaluation of survival data and two new rank order statistics arising in its consideration, Cancer Chemother. Rep. 50, 163-170 (1966). 8. Cox, D. R. Regression models and life tables, J. R. Stat. Sot. 34, 187-202 (1972). 9. Boyce, J., Fruchter, R. G., Nicastri, A. D. Ambiavagar, P. C. Reinis, M. S., and Nelson, J. N. Gynecol. Oncol. 12, 154-165 (1981). 10. West, R. R. Cervical cancer: Age at registration and age at death, Brit. J. Cancer 35, 236 (1977).
ONCOLOGY
30, 51-56 (1988)
Cervical Cancer Survival in Nijmegen Region, The Netherlands, 1970-l 985 YOLANDA VAN DER GRAAF, M.D., PH.D. ,* PETRONELLA G.M. PEER, GERHARD A. ZIELHUIS, M.Sc., PHD.,* AND PETER G. VOOIJS M.D., Departments
of *Social Medicine and fpathology, Geert Grooteplein Nijmegen, The Netherlands
M.Sc. ,* PH.DP
Zuid 24, 6525 Ga
Received December 29, 1986 Survival rates were computed for 359 women diagnosed with invasive cervical cancer between 1970 and 1985. The 5-year survival rate for the entire group was 67%. Survival was better in the period 1976-1980. Extension of the tumor and age at diagnosis were important prognostic factors. The effects of clinical stage, age at diagnosis, and year of diagnosis were studied simultaneously with the proportional hazards model. The hazard rate increased with increasing age and increasing clinical stage. Year of diagnosis had effect on survival only in case of IIB tumors. For the other clinical stages there was no significant effect on survival of year of diagnosis. 0 1988 Academic Press, Inc.
INTRODUCTION Cervical mortality rates are declining in several Western countries. [l]. In The Netherlands cervical cancer mortality has decreased since 1962. Since cervical cancer screening programs in The Netherlands did not start before 1976, cervical screening cannot be an explanation for this decline. A fall in the incidence of cervical cancer or a shift toward diagnosis of invasive cervical cancer in a less advanced stage seems the best explanation for this “natural” decline. However improvements of therapy with a better patient survival could also account at least for some of the decrease in mortality from cervical cancer. Ketting computed survival rates for all patients with invasive cancer of the uterine cervix treated in a large municipal hospital for the calendar period 1950-1979 [2]. He reported an improvement in therapeutic results with stage I, II, and III cervical cancer. He proposed the change in the treatment scheme in the period 1961-1969 as an explanation for the decrease in mortality. In this study we evaluated the survival of cervical cancer patients in the Nijmegen region for the calendar period 19701985 in order to investigate improvement in therapeutic results in the last 15 years. PATIENTS AND METHODS Population screening for cervical cancer started in 1976 in three pilot regions in The Netherlands. In these regions a quarter of the Dutch population is living. 51 00908258188 $1.50 Copyright 0 1988 by Academic Press, Inc. All rights of reproduction in any form reserved.
52
VAN
DER
GRAAF
ET AL.
All women aged 35 through 54 were invited every 3 years for a cervical smear. Apart from the screening program an increasing number of smears are being made by general practitioners and gynecologists. In the region of Nijmegen (about 750,000inhabitants) data on the incidence of invasive cervical cancer were available from the local cancer registry which has been kept since 1970 [3]. From the files of all pathology laboratories in the region all women were registered who had a histological diagnosis of microinvasive or invasive cancer of the uterine cervix. A histological diagnosis made in a pathology laboratory outside the region was not available for registration, but such cases were exceptions. Items registered were age at diagnosis, date of diagnosis, and clinical stage of the cervical cancer. Classification of the extension of the tumor is done according to the FIG0 classification (4). All diagnoses of cervical cytological smears made in the screening program as well as by general practitioners and gynecologists were added to the registry of histological diagnoses. All patients known to have cervical cancer were followed for vital status with the help of the registrar’s offices and the Central Bureau of Statistics. For a few cases this information was obtained directly from gynecologists and general practitioners. Since several patients were alive at the time the study was terminated and other patients were lost to follow up, survival curves were computed according to the technique of Kaplan-Meier [5]. Death from other causes is considered censoring. The standard errors of the estimated survival rates were computed according to the Greenwood formula (6). To test the equality of survival curves for different groups, the Mantel log rank test was used (7). In order to assess the effect of clinical stage of disease, age at diagnosis, and year at diagnosis simultaneously we used the proportional hazards model according to cox [8]. RESULTS During the study period 378 women were diagnosed with invasive cervical cancer. No follow-up data could be traced for 10 women and for 9 women the clinical stage of the tumor remained unknown. These women were left out of the analysis. Table 1 summarizes the clinical stage distribution according to age and point of time at diagnosis for the remaining 359 women. The Kaplan-Meier survival curve for the total group of women is displayed in Fig. 1. Five-year survival was calculated at 67.4% (standard error 2.6%). This plot illustrates that j-year survival is a good parameter for survival since only a few women die from cervical cancer after the fifth year. Table 2 shows the effect of extensions of the tumor, age, and calendar period on j-year survival. In general women with less advanced stages have a better survival. Exceptions are women with cervical cancers in clinical stage IIA. These women had a better survival then women with clinical stage IB tumors, although not significantly (Mantel log rank test, P = 0.52). All other pairwise comparisons yielded significant results with the log rank test. Survival curves for each stage are plotted in Fig. 2.
CERVICAL
CANCER
SURVIVAL
IN
TABLE CLINICAL
STAGE DISTRIBUTION
THE
53
NETHERLANDS
1
ACCORDING TO AGE AND CALENDAR
PERIOD
Clinical stage IA
IB
IIA
IIB
III
IV
Total
71(35%) 13(8%)
54(26%) 28(18%)
18(9%) 12(8%)
31(15%) 51(33%)
23(11%) 42(27%)
8(4%) 8(5%)
205 154
27(19%) 38(28%) 19(25%)
35(24%) 25118%) 22(29%)
13(9%) 14(10%) 3(4%)
41(28%) 28(20%) 13(17%)
23(16%) 27(20%) 15(19%)
6(4%) 5(4%) 5(6%)
145 137 77
Age <54 255 Calendar period 1970-1975 1976-1980 1981-1985
Survival rates differed also between age groups as is shown in Table 2. Survival was better in younger women. Although survival rates differed significantly for the three calendar periods, they do not show a trend. To study the effect of clinical stage, age at diagnosis, and year of diagnosis simultaneously, these three factors were incorporated into the proportional hazards model according to Cox. This multivariate regression model takes into account the correlation among the three factors. Because it was unlikely that for the same tumor stage, improvement of therapy differed for different ages and that the effect of age on survival was different for the various clinical stages, we did not take into account interaction of age with year of diagnosis and tumor stage. It could be, however, that therapy improved in the course of time only for some clinical stages. For that reason, age at diagnosis, clinical stage, year at diagnosis, and the interactions of clinical stage and year of diagnosis were entered in the model. The effect of year of diagnosis was significant only in clinical stage IIB and so in the final model age at diagnosis, clinical stage and the effect of year at diagnosis in clinical stage IIB were entered (Table 3). Increasing age and increasing clinical stage gave a significantly higher mortality. For stage IB and IIA, however, there was no significant difference in survival. The survival rate improved over the years for women with a stage IIB tumor. DISCUSSION The overall 5-year survival rate calculated in this study population is comparable with survival rates reported by others [2,9,10]. The “extension” (clinical stage) of the tumor seems the best predictor of survival. Less advanced clinical stageswith better survival were more frequently diagnosed in young women. Only part of these better survival rates in the youngest age groups can be explained by this phenomenon since after correcting for clinical stage the effect of age remained. Survival improved significantly during the last 15 years only for women with a stage IIB tumor. In the 1980s survival rates for these women became as high as for women with IB and IIA tumors, while in the period before 1980 the 5-year survival was calculated at 46%. This gain in
VAN
survival
DER GRAAF
ET AL.
__~--
80
60
50
I 48
I 24
I 96
I 72
I 120
I 144 follow
I 168
up
I 192 (months)
FIG. 1. Survival curve of women with cervical cancer. Region of Nijmegen, The Netherlands, 1970-1985.
TABLE 2 S-YEAR SURVIVAL RATES ACCORDING TO CLINICAL STAGE, AGE GROUP, AND CALENDAR PERIOD
Clinical stage (FIGO):
Age group:
IA
IB
IIA
IIB
III
IV
97%
77%
83%
55%
36%
11%
35-54
<35 91%
Calendar period:
55-70
72%
60%
>70 34%
1970-1975
1976-1980
1981-1985
62%
72%
67%
CERVICAL
CANCER
SURVIVAL
IN
THE
55
NETHERLANDS
survival
60
40
-7
-1
I L---
C------.
m
1
20
I L------------------X
Ip
1
1
I
I
12
24
36
48
I
60
/
72 follow-up
I
84 ( months
1
FIG. 2. Survival curves of women with cervical cancer according to clinical stage. Region of Nijmegen, The Netherlands, 1970-1985.
life expectancy must be the result of improvement in radiotherapeutic treatment since all women with a stage IIB tumor were treated with radiation therapy with curative intent. This finding is comparable with the results described by Ketting [2]. In that study the 5-year survival was 40% for stage IIB patients in the period 1961-1969 and 70% in the period 1970-1974 and comparable with the 5-year survival rate in stage IIA patients. The 5-year survival rate for cases with a IB tumor was somewhat lower in our study than that reported by Ketting [2]. In his study the 5-year survival for these cases was 84% in the period 1970-1974. Possibly this difference can be explained by the fact that the study populations were not comparable. Our study was population based and the women were treated in the regional hospitals and in one university hospital, while in the study of Ketting all women were treated in a center. The population screening led to a decrease of the incidence of invasive cervical cancer. In the period prior to the population screening program (1970-1976) the yearly incidence of invasive cervical cancer was 18.6 per 100,000 in women aged 35 through 54 years. During the first screening period the incidence was 21.5 per 100,000 women, in the second screening period 9.0 per 100,000 women and
56
VAN DER GRAAF ET AL. TABLE 3
REGRESSIONCOEFFICIENTSFOR CLINICAL STAGE, AGE AT DIAGNOSIS, AND THE EFFECT OF YEAR OF DIAGNOSIS IN STAGE IIB IN Cox’s PROPORTIONALHAZARDS MODEL
Variable Age at diagnosis (reference age 50) Year of diagnosis in stage IIB (reference 1975) Clinical stage (reference IB) IA IIA IIB III IV
Standard error
Quotient
P value (one sided)
0.0078
2.34
0.0097
-0.10
0.0467
2.20
0.0137
- 2.30 -0.33 0.65 1.26 2.20
0.7421 0.4635 0.2820 0.2762 0.3716
3.10 0.70 2.32 4.59 5.92
0.0009 0.2408 0.0102 0.0000 0.0000
Regression coefficient 0.018
in the third screening period 3.3 per 100,000 women. The incidence of invasive cancer among women older than 54 years, who were not included in the cervical cancer screening program, showed a decline 6 years after the start of the program. This can be explained by the increasing percentage of women in this age group who in previous years were eligible for the cervical cancer screening program and hence received treatment when a preinvasive lesion was discovered. In the course of 15 years a shift to more favorable clinical stages became evident. This shift to less advanced stages of cervical cancer at diagnosis and the improvement of therapy for women with a stage IIB tumor are responsible for a better overall survival. For the other clinical stages survival did not improve significantly during the last 15 years. ACKNOWLEDGMENT We are grateful to Jeep Hopstaken for collecting the data.
REFERENCES 1. Hill, G. B. Cancer of the uterus mortality trends since 1950, WHO Chron. 30, 188-193 (1976). 2. Ketting, B. W. Surgical treatment of invasive carcinoma of the uterine cervix, Thesis, University of Amsterdam, Rodopi, Amsterdam (1981). 3. van der Graaf, Y., Klinkhamer, P. J. J. M., and Vooijs, G. P. Effect of population screening for cancer of the uterine cervix in Nijmegen, The Netherlands, Preven. Med. 15, 682-690 4. Annual Report on the results of treatment in carcinoma of the uterus, vagina and ovary (H. L. Kottmeier, Ed.), Vol. 15, Stockholm (1971). 5. Kaplan, E. L., and Meier P. Non parametric estimation from incomplete observations. Amer. Stat. Assoc. J. 53, 457-481 (1958). 6. Lee, E. T. Statistical methods for survival data analysis, Univ. of Oklahoma Press, Oklahoma City, 1980. 7. Mantel N. Evaluation of survival data and two new rank order statistics arising in its consideration, Cancer Chemother. Rep. 50, 163-170 (1966). 8. Cox, D. R. Regression models and life tables, J. R. Stat. Sot. 34, 187-202 (1972). 9. Boyce, J., Fruchter, R. G., Nicastri, A. D. Ambiavagar, P. C. Reinis, M. S., and Nelson, J. N. Gynecol. Oncol. 12, 154-165 (1981). 10. West, R. R. Cervical cancer: Age at registration and age at death, Brit. J. Cancer 35, 236 (1977).