Department of Epidemiology and Biostatistics, Institute of Experimental and Clinical Medicine, and Estonian Cancer Registry, North Estonian Regional Hospital Foundation's Cancer Centre, Tallinn, Estonia
Estonian Cancer Registry, North Estonian Regional Hospital Foundation's Cancer Centre, Hiiu Str. 44, 11619 Tallinn, Estonia
During the past decade, the population of Estonia has undergone major political, social and economic changes. Estonia was a part of the USSR in 1940–1991, but its living standard was relatively high among the Soviet republics. After reestablishment of the national independence in 1991, Estonia became an open-market economy country, and the health care system was transformed from a centralised state-controlled system into a decentralised health insurance-based one. The Soviet health care system was funded from the state budget and claimed to provide health services free of charge to the whole population. The central authority, responsible for the operation of the system, was the Ministry of Health. According to the standard health insurance indicators, this health care system was one of the best equipped in the world, with a physician and bed supply greater than in any western country, but the quality of the services has never been expressed in terms of statistical indicators.1
The insurance-based health care system has been effective since 1991 and covers almost all medical services in Estonia. The scale of services is fixed in the price list that is regularly revised and approved by the Minister of Social Affairs. This system is mostly financed by employers, who pay from the salary fund a 33% social tax, from which 13% goes for the health insurance. The additional sources for health care are the state and municipal budgets and private out-of-pocket payments. The gross domestic product and the health expenditure are, however, considerably lower in Estonia compared to the more affluent European countries, e.g., the Nordic countries. In 1999, the total health expenditure per capita was 7-fold lower in Estonia than in Finland and 12-fold lower than in Norway.2
Cancer patients' survival is known to be strongly dependent on socioeconomic factors, as well as access to and quality of medical care. In Europe alone, there are substantial differences in survival rates of cancer patients between countries, which are clearly linked to health care inequalities.3, 4 In the past, survival rates in Estonia have been considerably lower than survival rates of cancer patients in other European countries, including Finland, a neighbouring country with an ethnically similar but economically more affluent population.5 However, previous publications exclusively reflect the survival experience of patients diagnosed in 1989 or earlier years.3, 5 The objective of our analysis was to assess the development of cancer patient survival in Estonia before, during and after the transition from a Soviet republic to an independent democratic country.
MATERIAL AND METHODS
Our analysis is based on data from the Estonian Cancer Registry (ECR), which is population based and covers the whole country (territory 45,216 km,2, population 1.4 million according to the 2000 census). Cancer registration in Estonia dates back to 1953, when compulsory notification of new cases of cancer was introduced in the former USSR. The ECR was founded in 1978, but reliable incidence data have been available since 1968, when cancer registration in Estonia became centralised.6, 7 Reporting of cancer cases is mandatory by the Decree, issued by the Ministry of Social Affairs, obligating the treating physicians and pathologists to notify the ECR of new cancer cases. The ECR follows the patients up to their death or emigration by linkage to the annual file of all death certificates issued in Estonia and to the Estonian Population Registry. During the entire period of operating, the basic data collection and follow-up procedures have been quite uniform in the ECR, ensuring the comparability of patients' survival rates between different time periods.6
We analysed trends in 5- and 10-year survival rates for patients with the 11 most common forms of cancer (lung, stomach, breast, colon and rectum, uterine cervix, prostate, uterine corpus, pancreas, ovary, kidney, urinary bladder) diagnosed in Estonia from 1969–1998. Nonmelanoma skin cancers were excluded from our study; in case of multiple cancers, only the first cancer was included. Patients diagnosed at all ages were included, but we excluded those reported to the ECR by the death certificate only (less than 1% overall) or first diagnosed by autopsy (about 4% overall). For this analysis, the follow-up of patients with respect to vital status was complete until the end of 1998.
To allow for timely detection of recent trends in long-term survival rates, the analyses were performed by calendar periods8, 9 rather than in the traditional manner by cohorts of patients diagnosed within certain years.10, 11, 12 The principle of period analysis has been described in detail elsewhere.8, 9 Briefly, period estimates exclusively reflect the survival experience of patients during some defined calendar period. This is achieved by left truncation of observations at the beginning of that period in addition to right censoring at its end. It has been shown both theoretically8, 9 and empirically13, 14, 15 that period analysis provides more up-to-date survival estimates than traditional (cohortwise) survival analysis.10, 11, 12 In particular, it has been demonstrated by extensive empirical evaluations that period estimates of long-term survival rates for some recent calendar period come very close to the survival rates actually experienced by patients diagnosed in that period, which can only be observed with a substantial delay by traditional methods of survival analysis.14
The principle of period analysis is illustrated in Table I for derivation of 10-year survival rates in 1994–1998, the latest period included in our analysis. For this period, survival experience during the first year following diagnosis is contributed by patients diagnosed from 1993–1998; survival experience during the second year following diagnosis is contributed by patients diagnosed from 1992–1997, and so on, until survival experience during the 10th year following diagnosis, which is contributed by patients diagnosed from 1984–1989. Overall, the survival experience of patients diagnosed from 1984–1998 is used, but the analysis is restricted to the survival experience in 1994–1998, which ensures a more up-to-date estimate of the 10-year survival than that obtained by traditional (cohortwise) methods. The latter would pertain to the survival experience during 1984–1998 of patients diagnosed in 1984–1988 or later years.
Table I. Period Analysis of 10-Year Survival for the 1994–1998 Period1
Only the survival experience in the bold black frame is included in the analysis. The numbers within the cells indicate the years of follow-up after diagnosis.
Period estimates of 5- and 10-year relative survival rates were derived and graphically displayed for 5-year calendar periods from 1974–1998. Throughout this article, we present relative rather than observed (absolute) 5- and 10-year survival rates of cancer patients. Relative survival rates are obtained by dividing observed survival rates by the expected survival rates of people with corresponding age and sex.12 The latter were derived from population life tables for Estonia according to an approach commonly known as Ederer II method (with appropriate adaptations for the period analysis approach).16 Relative survival rates reflect the “net mortality” due to cancer, and the 5- and 10-year survival trends shown in our article are therefore unaffected by mortality trends due to other causes of death. Period analysis of survival was performed by a recently developed special computer program.17 To facilitate interpretation of the survival trends, we provide the age-adjusted (to the world standard population) incidence and mortality rates per 100,000 person-years in Estonia for the same cancer sites and calendar periods, as included in the survival analyses.
The overall numbers of patients included in the analysis and the proportion of women by cancer site are shown in Table II, which also presents the mean age of patients at diagnosis and the age-adjusted incidence and mortality rates by cancer site and 5-year period of diagnosis from 1969–1998.
Table II. Number of Patients and Proportion of Women Included in the Analyses by Cancer Site, Mean Age of Patients at Diagnosis (Years), Age-Adjusted Incidence and Mortality Rates1 by Cancer Site and Calendar Period in Estonia, 1969–1998
Per 100,000 person-years, adjusted to the world standard population.
Total number of patients in 1969–1998 included in the survival analysis.
The most common malignancies were lung and stomach cancer, followed by breast and colorectal cancer. The vast majority of patients with lung and bladder cancer were men, whereas the proportion of women ranged from 44–57% for the other nongynaecological cancers.
Mean age at diagnosis varied from 56–60 years for patients with breast and cervical cancer to 70–72 years for patients with prostate cancer. For most cancer sites, the mean age at diagnosis was rather constant over time, but there was a moderate increase in mean age (by 2–3 years overall within the 30-year time span) among patients with cancers of the colon and rectum, breast, corpus uteri and ovary. A more substantial increase of mean age was observed among patients with kidney cancer (from about 58 years in 1969–1973 to about 63 years in 1994–1998).
Over the last 30 years, both the incidence of and mortality from lung cancer increased about 1.5 times. By contrast, a steady decrease was observed in the incidence and mortality in stomach cancer. The incidence of breast and colorectal cancer almost doubled, whereas the mortality increased less markedly. Both the incidence of and mortality from cervical cancer decreased by about 25%, but the major part of this decline occurred already prior to 1984. The greatest increase in incidence was seen for cancers of prostate and kidney, which were diagnosed about 3 times as often in 1994–1998 as in 1969–1973, while the increase in mortality was less pronounced. The major increase in incidence was observed also for bladder cancer, without considerable changes in mortality. For the other common cancer sites (uterine corpus, pancreas, ovary), the changes in incidence and mortality over time were modest.
Trends in 5-year and 10-year relative survival rates by cancer site and 5-year calendar period between 1974 and 1998 are shown in Figures 1–3. The calculation of the 10-year survival rates for the 1974–1978 period requires inclusion of survival experience of patients diagnosed before 1969; therefore the 10-year relative survival rates are given only for the 1979–1983 and later periods.
For lung cancer, 5-year relative survival rates were around 7–8%, and 10-year relative survival rates were about 5–6%, with very little change over time (Fig. 1). Somewhat more favourable survival rates were observed for patients with stomach cancer. Despite some increase over time, the 1994–1998 estimate of the 5-year relative survival was still below 20%, however. For patients with colorectal cancer, both the 5- and 10-year survival rates increased from about 30% in the earlier years to about 40% in 1994–1998. Prognosis of patients with pancreatic cancer is exceedingly poor, although there appeared to be some improvement in recent years.
Divergent trends were seen for patients with the most common gynaecological cancers (Fig. 2). For breast cancer, the 5-year relative survival rates increased from about 55% in 1974–1978 to about 65% in 1994–1998. The 10-year rates also increased over time but continued to be substantially lower than the 5-year rates for this form of cancer due to the relatively large number of late deaths. A modest increase in survival over time was also observed for patients with ovarian cancer, which seems to have been somewhat accelerated between the last 2 calendar periods. Nevertheless, the 5-year relative survival rate was still below 30% in 1994–1998. For cervical cancer, a downward trend in the 5- and 10-year relative survival rates was noted, which only stopped and partly reversed in the more recent years. Five- and 10-year relative survival rates around 70% and 65%, respectively, have been observed for patients with endometrial cancer, with no consistent trends over time.
By contrast, dramatic increases in survival rates over time have recently occurred for patients with common cancers of the urinary tract including prostate cancer (Fig. 3). From 1984–1988 and 1994–1998, the 5-year relative survival rate increased from about 25% to more than 50% for patients with kidney cancer, from slightly more than 30% to almost 60% for patients with bladder cancer, and from about 35% to about 60% for patients with prostate cancer. Major increases were also seen in the 10-year survival rates between these 2 time periods.
Our analyses revealed strongly divergent trends in patients' survival for common forms of cancer diagnosed in Estonia over the past 3 decades, the last of which is characterised by major political and socioeconomic upheavals.
Like elsewhere in the world, the prognosis was very poor and hardly improved with time for lung cancer and pancreatic cancer. A modest but steady improvement was observed in the prognosis of the other common gastrointestinal malignancies (stomach cancer and colorectal cancer) over time, suggesting increasing efficacy of care. Nevertheless, the 5- and 10-year period survival estimates for 1994–1998, which are likely to be good predictors of the survival rates of the patients actually diagnosed in that period,14, 15 are still lower than the corresponding cohort-based rates already achieved in the more affluent European countries for patients diagnosed in 1985–1989.3 Furthermore, the survival rates of the patients with colorectal cancer in Estonia continued to lag behind the corresponding recent estimates for patients in Finland (pertaining to the 1995–1997 period) by 10–15%.18 While the changes over time were of similar magnitude within both the neighbouring countries with an ethnically comparable population,19 the differences in the survival rates between the countries essentially remained constant.
Despite some steady overall improvement with time, the most recent 5- and 10-year relative survival rates of the patients with breast cancer and ovarian cancer continue to be about 20% lower than the corresponding rates of the patients in Finland.18 In particular, no major increase was seen for breast cancer in recent years. The 5-year relative survival rate of patients with breast cancer in Estonia in 1994–1998 was 63% and the 10-year relative survival rate was 53%, whereas the corresponding rates for breast cancer patients in Finland in 1995–1997 were 83% and 74%, respectively. It has been shown that the late diagnosis, particularly in elderly women, is the main reason for the lower survival rates of breast cancer in Estonia;20 however, also a delayed adherence to the modern treatment protocols could have contributed to the poorer outcome of the patients. Diagnosis of breast cancer is mainly symptom based in Estonia, and only a small proportion of the cases is detected by the opportunistic mammographic screening, introduced in some regions of the country.21 Yet, this programme does not correspond to the definition of an organised mass screening, according to which the target population should be adequately identified and covered.22 As revealed by the Eurocare-2 High Resolution Study, the proportion of small tumours (T1NoMo) is much lower in Estonia than in the European countries where organised mammographic screening is carried out.23
The differences from recent survival rates of Finnish cancer patients were somewhat less pronounced (about 10–15%) for cervical and endometrial cancer, but no tendency of improvement with time was observed for these cancers. The survival rates of the patients with cervical cancer even tended to decrease until the early 1990s. In Estonia, an organised screening programme for cervical cancer has never been introduced,24 and the age-standardised incidence rate per 100,000 women-years was 4-fold higher in Estonia than in Finland in 1997.25, 26 Nationwide cytologic screening was introduced in Finland in the late 1960s, and a substantial decrease in the incidence of and mortality from cervical cancer followed; on the other hand, this malignancy was the only one, for which a decrease in patients' survival has been observed in Finland over time, explained by the change in the composition of cases toward less favourable outcomes.19 Possibly, the stage distribution of the invasive cases has changed to the same direction in Estonia, where cervical cancer is diagnosed only by routine gynaecological practice, leading to the apparent decrease in survival rates.
By contrast, a tremendous increase in the 5- and 10-year relative survival rates (from 20–30%) was observed within just 1 decade from 1984–1988 and 1994–1998 for the common urologic cancers (prostate, kidney and bladder). The increase in survival coincided with the rapid increase in incidence, whereas the changes in mortality were only modest. Although the prognosis of the patients with these malignancies has also improved in other countries, including Finland,19 the magnitude of the recent improvement in Estonia has been extraordinary. As a result, the 10-year relative survival rate for prostate cancer in Estonia in 1994–1998 is very close to recent estimates for the patients in Finland.18 Compared to more developed industrial countries, the incidence of prostate cancer was low in Estonia until the early 1990s, since when a dramatic increase has been observed.6 The main reason is probably a change in the diagnostic activity. Like in other countries, the introduction of the PSA determination and the increasing frequency of transurethral resections have led to discovering a number of asymptomatic prostate cancers in Estonia as well. The increase occurred particularly in elderly men: the age-specific incidence among the persons aged 80 years or over was about twice as high in 1993–1997 compared to that in 1988–1992.6, 25 Inclusion of the incidentally diagnosed cases with their generally favourable stage distribution is likely to have improved the overall survival rates, while the prognosis may not have changed much for the patients with advanced disease. This assumption is also supported by the fact that the mortality from prostate cancer has not decreased in Estonia.
For kidney cancer, the period estimates of 5-year and 10-year relative survival rates in Estonia in 1994–1998 also come very close to the recent rates reported for Finland.18 A steady increase in the incidence of the kidney cancer has been observed in Estonia since the late 1960s and has continued during recent years.25 The mortality increased moderately until the early 1990s and leveled off thereafter. As elsewhere, the use of the modern imaging techniques, such as ultrasonography (USG) and computer tomography (CT) markedly improved the possibility of diagnosing small tumours in kidneys. Concerning the availability of the new diagnostic equipment, Estonia considerably lagged behind the more affluent countries: USG was introduced in routine diagnostic practice only since the late 1980s and CT scanners in the 1990s. The rapid increase in survival rates for kidney cancer in the 1990s may reflect an increasing proportion of small tumours as a result of the above mentioned diagnostic improvements.
Earlier diagnosis is probably also the main contributor to the improved outcomes for cancer of the urinary bladder. New diagnostic techniques, enabling more thorough investigation of the tumour tissue, became available in Estonia since the early 1990s. Furthermore, bladder cancer statistics have been a subject of debate because of the classification problems between invasive and noninvasive tumours, especially when the international comparability of both the incidence rates and the survival rates was concerned.3, 27 In our analyses, only invasive tumours were included. Still, due to changes in the diagnostic criteria, the borderline lesions could have been increasingly labeled as invasive ones by pathologists in the 1990s. In 1988–1992, the highest age-specific incidence rate in Estonia was observed for people aged 75–79 years, and the rates dropped for people 80 years and over.28 In recent years, an increase was observed in the age-specific incidence among the people at advanced age,25 which could again result from changes in diagnostic activity.
Our study does not allow direct comparison of the effectiveness of the health care systems in Estonia before and after the reindependence. For a number of cancer sites such as stomach, colon and rectum, breast and ovary, an upward trend in patient survival was observable since the late 1970s and continued until the late 1990s. Although the extent and speed of the changes varied, as did the survival rates themselves, it is difficult to attribute these trends to the recent social and economic changes or to the health care reforms. Nevertheless, our results point to improved detection and treatment of cancer over time. The dramatic progress in survival rates for the urologic cancers (prostate, kidney and bladder) probably reflects recent breakthroughs in the diagnostic and, possibly, treatment methods in Estonia. The improved training of physicians (including the training abroad, which became largely available after the reindependence) may also have contributed to these major developments.
In conclusion, the period analysis presented in our article provides up-to-date estimates of the 5-year and 10-year survival rates of patients diagnosed with cancer in Estonia in the 1990s, when the country underwent the transition from a closed, totalitarian society to a democratic one with an open-market economy. The period estimates to some extent revise the “pessimistic” picture previously obtained from traditional, cohortwise survival studies, which were exclusively based on the data of patients diagnosed until 1989. The survival rates generally improved in Estonia by the end of the 20th century. Nevertheless, patients with the most common forms of cancers still experience poorer outcomes than the patients in more affluent countries, such as Finland. The survival trends observed in Estonia follow, although with a considerable delay, the trends observed in developed industrial countries 1–2 decades earlier. Despite the overall improvement, however, major efforts in delivering modern cancer care to the population of Estonia will be required to close the gap that continues to exist between the survival rates of cancer patients in this country and other European countries.
The authors gratefully acknowledge Mrs. P. Härmaorg for preparing the datafiles for analyses and calculating the incidence and mortality rates, and Mrs. H. Talvoja for her editorial assistance. We also thank 2 referees for their helpful comments on the manuscript.