Education, survival, and avoidable deaths in Lithuanian cancer patients, 2001–2009

Abstract

Background Our aim in this study is to provide a systematic assessment of the site-specific cancer survival rates of patients with different educational levels, using population-based census-linked registry data covering the entire population of Lithuania. Material and methods The study is based on the linkage between all records of the 2001 population census and all records from Lithuanian Cancer Registry (cancer incidence) and Statistics Lithuania (deaths) for the period between 6 April 2001 and 31 December 2009. Results For the vast majority of cancer sites we found an inverse gradient in survival, with the worst survival indicators in the lowest educational group. We estimated that 18.6% of the deaths in Lithuanian cancer patients could have potentially been postponed, if all the patients had the same cancer mortality as the patients with the highest educational level. Conclusion Our findings offer a warning that although the survival rates of cancer patients are improving, this progress hides disparities between different groups of patients.

In recent decades, substantial progress has been made in developing and implementing effective cancer prevention strategies, early detection interventions, and cancer treatments. This progress has resulted in an overall increase in survival rates for many cancer sites. Since 1990, the EUROCARE project has been analyzing data from European population-based cancer registries to determine whether there are differences in cancer patient survival across Europe. The most recent EUROCARE study reported that cancer survival rates are highest in northern, central, and southern Europe; and that survival rates are generally low and below the European mean in most countries of eastern Europe [1].

Much less is known about the differences in site-specific cancer survival by socio-economic status and the evidence on this topic is especially scarce for the eastern European region. There are relatively few population level studies that have looked at differences in the survival rates of cancer patients by socio-economic status, while also taking into account the specific cancer type. Cancer registry-based studies on socio-economic inequalities in cancer survival have been conducted in England and Wales [2], Scotland [3], Denmark [4], Finland [5], Germany [6], Italy [7], and the Netherlands [8]. With the exception of a few smaller-scale medical survey datasets (which are not representative of the entire population), there is little information for eastern European countries that would allow us to examine the relationship between educational level and survival among patients with specific types of cancer. Yet an investigation of this issue is particularly interesting because, compared to the rest of Europe, eastern European countries generally have lower life expectancy levels, higher overall mortality rates, and much lower levels of health care spending.

Despite having recently made progress in reducing overall mortality, Lithuania continues to lag behind the rest of the European Union in life expectancy at birth, especially among men: in Lithuania, life expectancy among men is 7–10 years lower than in the majority of western and even central European countries. Although the survival rates of cancer patients in Lithuania have recently been improving, the survival rates for most cancer sites in Lithuania continue to be below the European average, and are comparable to those of other countries of eastern Europe [1,9].

To our knowledge, socio-economic inequalities in site-specific cancer survival have so far been studied only in low-mortality countries with high (relative to the European average) survival rates. In Lithuania, existing census-linked cancer registry data has only been used to study socio-economic inequalities in cancer incidence and mortality. Our aim in this study is to provide for the first time in the eastern European region a systematic assessment of the site-specific cancer survival rates of patients with different educational levels who were diagnosed with cancer during the period 2001–2009, using population-based census-linked registry data covering the entire population of Lithuania. The public health impact of the observed differentials is assessed by estimating a number of avoidable cancer deaths that could be postponed under a hypothetical scenario in which the relative survival rates in all educational groups are the same as in the most affluent group.

Material and methods

Dataset

The study is based on a census-linked dataset covering entire population of Lithuania. The dataset includes all of the records of the 2001 population census, all of the records of the Lithuanian Cancer Registry (cancer incidence), and all of the death and emigration records of Statistics Lithuania for the period of 6 April 2001–31 December 2009. Linkages between the census, emigration, death, and Cancer Registry records were implemented using personal identification numbers as unique identifiers of individuals. For individuals who died or emigrated, the exposure time was censored at the date of death or emigration. All of the linkage procedures were implemented by employees of Statistics Lithuania, who have permission to work with individual level data. Only aggregated data in a frequency table format that combines aggregated cancer cases, deaths, and population exposures for every possible combination of relevant socio-demographic and epidemiological variables were provided for this study.

Almost all cases of cancer first diagnosed between 6 April 2001 (census date) and 31 December 2009 were successfully linked to the census data. Of the eligible cancer cases, 0.20% had to be excluded because the patients were not found in the 2001 population census. Only census-linked registry records were used in the analysis. In total, 111 703 newly diagnosed cancer cases were included in the final dataset. For this study, we chose 22 common cancers representing 81.7% of all cancer cases observed during the period of observation.

The population under study includes all of the permanent residents of Lithuania aged 30–74 years on the census day (6 April 2001). For both new cancer patients and survivors without cancer, information about education was taken from the 2001 census records. Education was classified according to three broad categories: higher education (at least 14 years of schooling), secondary education (10–13 years of schooling), and lower than secondary education (up to nine years of schooling). Based on our earlier experiences and international recommendations, individuals with unknown education (0.48% of the study population) were included in the lowest educational group.

The follow-up started at the date of the first cancer diagnosis and ended at the date of emigration or death, or on 31 December 2009 (the end of the observation period). We excluded cases in which the patient survived less than one month (e.g. if the patient died in the month of diagnosis) or in which the patient’s death was registered using only the death certificate (DCO) or autopsy information (Table 1).

Table 1. Study population (total and excluded cases of diagnosed cancer).

	Higher			Secondary			Lower than secondary
Site	Total	Excluded	%	Total	Excluded	%	Total	Excluded	%
Esophagus	86	11	12.8	543	95	17.5	635	144	22.7
Stomach	733	73	10.0	2459	326	13.3	3148	560	17.8
Colon & rectum	1456	86	5.9	3800	295	7.8	4388	577	13.1
Liver	125	35	28.0	392	112	28.6	526	182	34.6
Gallbladder	96	14	14.6	257	46	17.9	352	68	19.3
Pancreas	394	70	17.8	1193	256	21.5	1355	372	27.5
Larynx	103	3	2.9	619	29	4.7	770	65	8.4
Lung	823	130	15.8	3827	687	18.0	6360	1381	21.7
Skin melanoma	461	4	0.9	777	16	2.1	421	18	4.3
Breast	2160	28	1.3	5123	92	1.8	2753	160	5.8
Cervix uteri	460	4	0.9	2191	56	2.6	905	74	8.2
Corpus uteri	712	6	0.8	1847	42	2.3	1361	54	4.0
Ovary	495	28	5.7	1396	79	5.7	964	150	15.6
Prostate	3532	47	1.3	6323	110	1.7	7826	271	3.5
Kidney	783	34	4.3	2073	137	6.6	1801	220	12.2
Urinary bladder	304	9	3.0	1000	41	4.1	1344	115	8.6
Brain	297	43	14.5	837	119	14.2	616	141	22.9
Thyroid	532	4	0.8	1341	11	0.8	471	23	4.9
Hodgkin’s lymphoma	53	1	1.9	162	4	2.5	68	13	19.1
Non-Hodgkin’s lymphoma	347	28	8.1	803	51	6.4	758	88	11.6
Multiple myeloma	202	14	6.9	395	42	10.6	434	49	11.3
Leukemia	381	59	15.5	1026	145	14.1	1108	231	20.8

Methods

In order to calculate the relative survival rates, sex-specific life tables for the general population by each educational category were calculated using the same census-linked dataset; following Chiang’s methodology [10]. In the next step, we calculated the cumulative five-year survival rates for the period 2001–2009. We estimated relative survival to remove the background mortality (due to causes other than cancer) which can vary widely by education group and can bias cancer survival comparisons. Relative survival is defined as the ratio of the observed survival in patients with cancer to the expected survival in the corresponding group in the general population (calculated on the basis of the census-linked dataset covering entire population). Relative survival was estimated using the Ederer II method from education-specific life tables stratified by age, sex and calendar year. Because age structures are not homogeneous across educational groups, we applied a direct method of age-standardization using the International Cancer Survival Standards (ICSS) [11]. STATA statistical software was used to calculate relative survival and avoidable deaths, using the freely available “srtr” command [12].

To ensure that we had an adequate follow-up period, we used only cancers diagnosed between 2001 and 2004 for the estimation of excess of deaths that could be avoided up to five years after diagnosis. The excess and avoidable deaths were calculated according to the method described by Ellis et al. [2].

Results

After the exclusions described in the above section and in Table 1, there were 82 805 patients included in the final dataset. The age-standardised five-year relative survival rates for males and females are shown in Tables 2 and 3.

Table 2. Five-year relative survival of male cancer patients diagnosed in Lithuania in 2001–2009, by site and educational level.

	Higher		Secondary		Lower than secondary		Absolute difference, %
Site	RS	SE	RS	SE	RS	SE	Secondary vs. higher	Lower than secondary vs. higher
Esophagus	17.2	6.4	6.6	1.6	4.9	1.3	10.5	12.3
Stomach	29.6	3.1	29.6	1.6	19.2	1.4	0.0	10.4
Colon & rectum	56.3	2.6	50.2	1.7	45.8	1.7	6.1	10.5
Liver	8.5	6.9	15.0	3.7	10.1	3.0	−6.6	−1.6
Gallbladder	26.6	9.6	20.2	7.1	26.2	6.2	6.4	0.4
Pancreas	9.7	2.8	5.0	1.2	3.9	1.2	4.7	5.8
Larynx	68.4	6.8	51.2	2.9	45.1	2.8	17.2	23.3
Lung	11.6	1.8	11.3	0.8	7.7	0.5	0.3	3.9
Skin melanoma	68.1	5.0	55.3	4.2	55.7	7.1	12.8	12.4
Prostate	97.2	1.1	91.6	1.1	86.1	1.1	5.5	11.0
Kidney	76.3	2.7	64.4	2.0	50.4	2.5	11.9	25.9
Urinary bladder	59.5	4.6	63.8	2.5	55.7	2.5	−4.4	3.8
Brain	18.5	4.3	19.2	2.6	10.5	2.6	−0.7	8.0
Thyroid	95.6	4.5	94.0	3.6	81.9	7.1	1.6	13.8
Hodgkin’s lymphoma	86.2	7.8	70.1	7.5	41.1	12.4	16.1	45.1
Non-Hodgkin’s lymphoma	62.8	5.4	54.6	3.6	40.1	4.1	8.2	22.8
Multiple myeloma	47.0	7.3	34.3	5.7	26.3	5.2	12.7	20.7
Leukemia	59.1	4.9	50.1	3.2	38.6	3.4	9.0	20.5

RS, relative survival; SE, standard error.

Table 3. Five-year relative survival of female cancer patients diagnosed in Lithuania in 2001–2009, by site and educational level.

	Higher		Secondary		Lower than secondary		RS difference (vs higher), %
Site	RS	SE	RS	SE	RS	SE	Secondary	Lower than secondary
Esophagus	16.5	11.2	16.0	6.7	12.0	5.1	0.6	4.5
Stomach	32.9	3.5	31.6	1.9	29.6	1.9	1.2	3.3
Colon & rectum	59.3	2.5	52.0	1.6	47.3	1.5	7.2	11.9
Liver	17.6	7.9	25.9	5.2	10.3	3.7	−8.3	7.3
Gallbladder	26.5	7.8	16.2	3.9	17.2	3.3	10.3	9.3
Pancreas	8.4	3.8	8.2	1.8	5.5	1.3	0.2	2.9
Lung	23.3	4.3	14.1	2.0	11.2	1.7	9.2	12.1
Skin melanoma	84.4	2.9	74.7	2.7	69.5	4.3	9.6	14.9
Breast	80.2	1.2	73.6	0.9	64.8	1.4	6.6	15.3
Cervix uteri	74.5	2.4	65.3	1.2	50.0	2.1	9.2	24.5
Corpus uteri	81.6	2.0	83.9	1.2	75.2	1.7	−2.3	6.4
Ovary	51.2	3.1	46.6	1.8	29.3	2.0	4.6	21.9
Kidney	77.4	3.4	75.3	2.1	61.2	2.5	2.1	16.3
Urinary bladder	73.9	6.9	77.3	3.8	56.5	4.4	−3.4	17.4
Brain	19.7	4.0	24.5	2.7	4.6	1.8	−4.8	15.1
Thyroid	97.1	1.4	98.6	0.8	92.9	2.5	−1.5	4.2
Hodgkin’s lymphoma	85.9	8.7	81.8	5.0	59.9	11.5	4.1	26.0
Non-Hodgkin’s lymphoma	69.0	4.4	65.9	3.2	48.8	3.5	3.1	20.2
Multiple myeloma	41.7	6.2	33.3	4.8	26.8	3.8	8.4	14.9
Leukemia	58.7	5.6	52.8	3.1	46.5	3.3	5.8	12.1

RS, relative survival; SE, standard error.

We found systematically lower survival rates among cancer patients with lower educational levels. Among men, relative survival rates were significantly lower in the lowest educational group for 19 of the 20 cancer sites analyzed. Among women, we found the same inverse educational gradient in cancer survival for all 21 cancer sites under study.

Among men, the absolute difference in survival rates between the highest and the lowest educational groups was exceptionally high for Hodgkin’s lymphoma (45%). Substantial gaps (above 20%) were also found for kidney cancer, larynx cancer, non-Hodgkin’s lymphoma, multiple myeloma, and leukemia (Table 2). Among women, the largest difference in relative survival rates was for Hodgkin’s lymphoma (26%), though gaps of more than 20% were also found for cervical and ovarian cancers and non-Hodgkin’s lymphoma. For both sexes, the absolute differences in the relative survival estimates for cancer sites with good and poor prognoses were much less pronounced. The absolute difference for cancers with poor prognoses ranged from 2.9% (pancreas) to 15.1% (brain) among women, and from 3.9% (lung) to 12.3% (esophagus) among men.

The differences in the survival rates of cancer patients based on educational level have important public health implications. Under a hypothetical scenario in which all of the educational groups had the same survival rates as those of the highest educational group, 2729 of the total number of 14 689 cancer-related excess deaths would have been avoided (postponed) during the 2001–2004 period (Table 4). The largest contributors to the total number of avoidable cancer deaths were prostate, breast, lung, colorectal, and stomach cancers (the most common cancers in the study population). Among patients diagnosed with those cancers, 1481 deaths would have been avoided (postponed) within five years of the diagnosis (under the same hypothetical scenario of the elimination of inequality).

Table IV. Number of cancer patients, number of excess deaths and avoidable deaths in Lithuania five years after diagnosis (hypothetical situation in which all patients have the same relative survival as those in the highest educational category).

Site	Number of patients	Excess deaths	Avoidable deaths	%
Esophagus	397	359	106	29.5
Stomach	2407	1710	214	12.5
Colon & rectum	3550	1723	235	13.6
Liver	285	243	35	14.5
Gallbladder	260	206	36	17.4
Pancreas	925	857	5	0.6
Larynx	587	282	183	65.0
Lung	3931	3451	295	8.5
Skin melanoma	641	192	59	30.5
Breast	3817	1068	298	27.9
Cervix uteri	1519	592	170	28.8
Corpus uteri	1567	316	33	10.5
Ovary	1114	627	102	16.3
Prostate	3926	623	439	70.6
Kidney	1028	381	156	41.0
Urinary bladder	1600	552	117	21.2
Brain	362	273	6	2.3
Thyroid	428	263	28	10.5
Hodgkin’s lymphoma	836	59	25	42.5
Non-Hodgkin’s lymphoma	399	167	52	31.4
Multiple myeloma	510	253	70	27.7
Leukemia	698	399	48	12.0
Total	31043	14689	2729	18.6

The highest percentage of avoidable excess deaths was found for prostate cancer (70.6%), which was also shown to be the largest contributor to the total number of avoidable deaths. The percentage of avoidable excess deaths was very low for lung cancer (8.5%). Similarly low percentages of avoidable excess deaths were found for other lethal cancers, such as brain and pancreas cancers (2.3% and 0.6%, respectively).

For cancers with good prognoses (melanoma, kidney cancer, Hodgkin’s and non-Hodgkin’s lymphomas), the shares of avoidable deaths were large (over 30% on average), even though the absolute numbers of avoidable deaths were small. Moreover, while the overall survival rates for these cancers were high, the survival rates differed substantially by education group.

Discussion

To our knowledge, this is the first register-based population level study in eastern Europe that provides evidence on cancer survival differences by education. The study is based on census-linked cancer registry data covering more than 82 000 cancer patients. The use of these data, which are unique in the region, ensures that our results are consistent and internationally comparable. Both the numbers of missing cases of education and the percentage of census-unlinked deaths in the data are negligible, and thus cannot have a significant influence on the results. A relative survival approach was used to assess and compare the survival rates of cancer patients of different education groups. One of the strengths of this study is that the relative survival rates were calculated using the expected survival rates derived for each educational group. This approach, which was chosen based on prior methodological recommendations, allowed us to account for the much higher overall mortality levels among the lower educational groups, and to avoid overestimating the survival gap between the different educational groups [13].

The patterns of educational inequalities in cancer survival by educational level in Lithuania are generally similar to those observed in other European countries [2–8]. For the vast majority of cancer sites, we found an inverse gradient in survival, with the worst survival indicators in the lowest education group. However, we also found that for some cancer sites – such as cervical and ovarian cancers for women, larynx and kidney cancers for men, and Hodgkin’s and non-Hodgkin’s lymphomas for both sexes – the survival differences are higher than 20% between educational groups. This finding is particularly striking given that the overall survival rates in Lithuania are already among the worst in Europe. This result therefore highlights the exceptionally unfavorable situations of lower educated patients diagnosed with these cancers.

One of the important contributions of the present study is that it provides estimates of the numbers of deaths that could be avoided (postponed) if the educational inequalities in cancer survival in Lithuania were eliminated. These estimates give us a better understanding of the impact of cancer on the overall public health burden. Our results suggest that the proportion of avoidable cancer deaths in Lithuania is higher than the shares reported in England (10% three years after diagnosis) and Finland (4–7% five years after diagnosis) [2,5].

In the most recent review three broad groups of factors (relating to the tumor, the patient and the health care system) influencing cancer patients survival have been extensively discussed [14]. For a long time tumor-related determinants, such as differences in the stage of diseases at diagnosis (being one of the most important clinical prognostic factors), or variations in biological characteristics of the tumor were one of the most widely used explanations. Patient-related determinants contributing to the educational differences in cancer survival include health behaviors, presence of co-morbidity, and psychosocial factors [14,15]. Many studies revealing a general mortality disadvantage of lower socio-economic groups point to the role of adverse material circumstances, a lack of personal control over life events and health, unhealthy behaviors, a higher level of exposure to psychosocial stress, and poor childhood conditions [16]. Lithuanian health surveys systematically point to higher rates of smoking, alcohol consumption, and unhealthy food among lower educated people [17,18]. The cancer registry-based study found a significantly increased risk of suicide among the lower educated cancer patients in Lithuania [19]. This may indicate a higher prevalence of psychosocial problems in this group of patients. Although it has been argued that a part of socio-economic differences in cancer survival may depend on the presence of physical co-morbidity (the presence of other diseases), the existing evidence is mixed and varies depending on cancer site and country [14]. For example, a Dutch study found that co-morbidity is higher among deprived patients with the cancers of breast and lung whereas such pattern is not typical for the patients with the stomach or prostate cancers [20]. A recently published Danish study also concluded that socio-economic differences in survival after lung cancer cannot be fully explained by the corresponding differences in the stage, treatment, and co-morbidity [21].

Some studies have shown that determinants related to health care, such as unequal access to treatment, are important factors in the differences in the survival rates of cancer patients by socio-economic status [14]. For example, disadvantages in receiving timely treatment or breast, colorectal or lung cancer surgery among less educated or poor (with lower income) individuals or people living in socio-economically deprived areas have been reported in Italy, England, and Denmark [21–24]. Evidence from Denmark suggests that substantial socio-economic differences also exist in cancer rehabilitation and this may contribute to the observed disparities in cancer survival [25]. Also, the participation in screening activities seems to be also less common in lower socio-economic groups [26].

The general factors that affect the overall health of the entire population may also influence the survival of cancer patients, such as the financing of health care and the technological and human resource investments in the health care sector. In the late 2000s, Lithuania was found to be the EU country with the lowest levels of overall spending per capita on cancer treatments and cancer drugs [27]. This indicates that the health care system in Lithuania has general structural problems, which are obviously associated with disadvantages in the treatment of a range of diseases in addition to cancer (e.g. cardiovascular mortality) and in life expectancy. It is important to stress that this generally unfavorable situation hides both a striking degree of variability in mortality across socio-economic groups, and extremely high mortality among some disadvantaged groups. For example, it has been shown that the life expectancy gap between men with the highest and lowest levels of education exceeds 11 years [28]. Therefore, it is likely that at least a portion of the observed difference in the survival rates of cancer patients is explained by many of the same factors that contribute to the huge overall health disadvantage of lower educated men and women in Lithuania.

The results of this study are important for clinicians providing general guidelines about specifics of certain groups of patients [29]. It has been suggested that the population level evidence about the advanced survival of highly educated patients may be used as a general benchmark for the remaining patient groups [24]. However, more comprehensive data that take into account more detailed patient and tumor characteristics are needed in order to correctly assess survival of individual patients which may substantial differ within large educational groups.

In conclusion, this study found the very large inequalities in the survival rates of cancer patients in Lithuania. Our findings also offer a warning that although overall survival rates are improving in Lithuania, this progress may be very uneven across different patient groups. The results of this study indicate that in Lithuania less educated group of cancer patients is more vulnerable and require special attention. These gaps should be addressed through appropriate inter-sectorial policies and measures aimed specifically at lower educated patients.

Disclosure statement

The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.