**4. Analysis**

In our analysis, we examine the impact of the lists of causes of death on amenable mortality by country, sex and cause of death. We compare the results of amenable mortality across the European Union (EU) countries calculated by the four lists of causes of death. Then, we are interested whether the two latest developed lists (AMIEHS and EUROSTAT) have a statistically significant impact on amenable mortality in Slovakia identifying the most influential group of diseases.

#### **4.1. Data and methods**

This section introduces what kind of dataset and methods are applied on the estimation of age‐standardised amenable death rates when comparing the EU countries. It also includes information how significances of the results have been tested.

#### *4.1.1. Data*

by Delphi method; unfortunately, those results are not disseminated. The final EUROSTAT Satellite List defining causes of death considered as amenable or preventable is available at Eurostat web page [24]. We present the list of causes considered to be amenable in **Table 2**. As one should notice, the development of concept of avoidable mortality has been considerably influenced by the evidence from clinical research studies or consultation that has confirmed the impact of health care or public health interventions on declining mortality. However, a considered time period has played an important role in creating the unique list of selected diseases, because medical knowledge and technology have advanced over time what subsequently has an impact on inclusion or exclusion criteria by which a list of amenable or preventable causes of death is made. Therefore, the lists of causes of death amenable to health

Although avoidable mortality has been investigated for the last four decades, there is still small consensus among researchers about how to define it. Last precise definitions of the concept are presented by the Office for National Statistics in England [25]. Following definitions

Avoidable deaths are all those defined as preventable, amenable (treatable) or both, where each death is counted only once; where a cause of death is both preventable and amenable, all deaths from that cause are counted in both categories when they are presented separately.

A death is amenable (treatable) if, in the light of medical knowledge and technology at the time of death, all or most deaths from that cause (subject to age limits if appropriate) could be

A death is preventable if, in the light of understanding of the determinants of health at time of death, all or most deaths from that cause (subject to age limits if appropriate) could be

In our analysis, we examine the impact of the lists of causes of death on amenable mortality by country, sex and cause of death. We compare the results of amenable mortality across the European Union (EU) countries calculated by the four lists of causes of death. Then, we are

care need to be regularly updated in relation to current medical practice.

were developed through an iterative public consultation running in 2015.

**3.3. Office for national statistics in England**

*3.3.1. Avoidable mortality*

80 Advances in Health Management

*3.3.2. Amenable mortality*

*3.3.3. Preventable mortality*

**4. Analysis**

avoided through good quality healthcare.

avoided by public health interventions in the broadest sense.

Our main source of mortality data is the raw data files of the WHO Mortality Database, where the causes of death are coded using the ICD‐10 classification at fourth digit level by five‐year age groups. We conduct analysis on data from 2014, as it is the latest available time point. The data in the required structure for calculation of amenable mortality are available for 19 EU countries, while other EU countries do not meet the requirements of this analysis due to data incompleteness at some age groups. We select causes of death that are proposed by the Nolte and McKee, Tobias and Yeh, AMIEHS, EUROSTAT's list regardless to the age limit. Statistical database of the United Nations Economic Commission for Europe is the main source for data on mid‐year population at the age groups. For comparison of mortality across EU countries, we adopt the European standard population by age groups according to the last revision in 2012, proceeding in 2013 [26].

#### *4.1.2. Methods*

We estimate age‐standardised amenable death rates per 100,000 population by the direct method of standardisation to overcome an effect from variations in the age and sex structure across countries. First, the age and sex‐specific death rates for the given causes of death are calculated in each examined country. Second, the age‐specific death rate and the European standard population for each age interval are multiplied, and these results are summed. Finally, this sum is divided by the total standard population, in our case 100,000, to calculate the age‐standardised death rate [27].

Two directly standardised rates calculated by the same standard population can be compared, and differences tested for statistical significance. To determine an association of countries' rank order according to the standardised death rates between the lists each other, we run a Spearman rank‐order correlation with statistical significance tests. Probability values are computed from a *t*‐distribution with N‐2 degrees of freedom.

To find out whether age‐standardised rates of amenable mortality based on the two lists are significantly different by sex and causes of death in Slovakia, we calculate 95% confidence intervals that are equivalent to statistical tests. As a general rule, a difference is statistically significance if a confidence interval around rate non‐overlap with the interval around another [28]. Calculations are made using statistical software R Studio.

#### **4.2. Between‐list differences of amenable mortality across the European Union countries**

This section compares the results of age standardised death rates across the European Union countries based on data from 2014 using the four evolutionarily most recent selections of amenable diagnoses. We tested the six null hypothesis statements (H0) against the six alternative hypotheses (H1):


**Table 3** reports the Spearman's rank correlation matrix with a statistical significance of correlation coefficients. All calculated probability values achieved a value of *p* < 0.001, what means that we can reject the null hypothesis. In other words, despite any concept of amenable mortality applied, there is a significant very strong positive correlation of the standardised death rates. Generally, the Spearman's correlation test calculated on standardised death rates of amenable causes using the Nolte and McKee, Tobias and Yeh, AMIEHS or EUROSTAT's concepts, shows that the *rank order of countries does not change significantly*.

Assessment of Avoidable Mortality Concepts in the European Union Countries, Their Benefits and Limitations http://dx.doi.org/10.5772/67818 83


*Note:* Probability values computed from a t distribution with N‐2 degrees of freedom. *N* = 19. *Source*: Own calculation using R Studio.

**4.2. Between‐list differences of amenable mortality across the European Union countries**

hypotheses (H1):

82 Advances in Health Management

This section compares the results of age standardised death rates across the European Union countries based on data from 2014 using the four evolutionarily most recent selections of amenable diagnoses. We tested the six null hypothesis statements (H0) against the six alternative

• H0: There is no association between the standardised death rates calculated by Nolte and

• H1: There is an association between the standardised death rates calculated by Nolte and

• H0: There is no association between the standardised death rates calculated by Nolte and

• H1: There is an association between the standardised death rates calculated by Nolte and

• H0: There is no association between the standardised death rates calculated by Nolte and

• H1: There is an association between the standardised death rates calculated by Nolte and

• H0: There is no association between the standardised death rates calculated by Tobias and

• H1: There is an association between the standardised death rates calculated by Tobias and

• H0: There is no association between the standardised death rates calculated by Tobias and

• H1: There is an association between the standardised death rates calculated by Tobias and

• H0: There is no association between the standardised death rates calculated by AMIEHS's

• H1: There is an association between the standardised death rates calculated by AMIEHS's

**Table 3** reports the Spearman's rank correlation matrix with a statistical significance of correlation coefficients. All calculated probability values achieved a value of *p* < 0.001, what means that we can reject the null hypothesis. In other words, despite any concept of amenable mortality applied, there is a significant very strong positive correlation of the standardised death rates. Generally, the Spearman's correlation test calculated on standardised death rates of amenable causes using the Nolte and McKee, Tobias and Yeh, AMIEHS or EUROSTAT's con-

McKee's list and the standardised death rates calculated by Tobias and Yeh's list.

McKee's list and the standardised death rates calculated by Tobias and Yeh's list.

McKee's list and the standardised death rates calculated by AMIEHS's list.

McKee's list and the standardised death rates calculated by AMIEHS's list.

McKee's list and the standardised death rates calculated by EUROSTAT's list.

McKee's list and the standardised death rates calculated by EUROSTAT's list.

Yeh's list and the standardised death rates calculated by AMIEHS's list.

Yeh's list and the standardised death rates calculated by AMIEHS's list.

Yeh's list and the standardised death rates calculated by EUROSTAT's list.

Yeh's list and the standardised death rates calculated by EUROSTAT's list.

list and the standardised death rates calculated by EUROSTAT's list.

list and the standardised death rates calculated by EUROSTAT's list.

cepts, shows that the *rank order of countries does not change significantly*.

**Table 3.** Spearman's rank correlation matrix with *p*‐values calculated for standardised death rates (sdr) by country based on the four lists of amenable causes, 2014.

These results are depicted in **Figure 1**. The four lists provide different levels of amenable mortality rates for countries; however, the rank order of countries is very similar. In 2014, France accounted for the best results of amenable mortality obtained from the all examined lists, ranged from 61 to 79 deaths per 100,000 population. On the other hand, the worst rate was recorded in Romania, 275 per 100,000 calculated by Nolte and McKee's list, as well as an average of 309 deaths per 100,000 in Latvia estimated by three remaining lists.

Generally, the standardised death rates for EU‐19 calculated by Eurostat's list were 40.5% higher than rates calculated by Nolte and McKee's list. On the other hand, the rates calculated according to the lists of Tobias and Yeh or AMIEHS were nearly the same, 161 per 100,000, 162 per 100,000, respectively. Using the Nolte and McKee's list, the amenable mortality rates for EU‐19 reached the lowest value, 128 deaths per 100,000 population. The standard deviations (not shown in this document) expressing the rate of variability of standardised amenable death rates between lists, gained the highest values in Eastern European countries (Latvia, Lithuania, Slovakia, Hungary, Romania, the Czech Republic, Poland), along with Denmark, Estonia, the United Kingdom, Croatia, had still standard deviations above the average of EU‐19. A gradual decline of the variation in amenable mortality rates, below an average of EU‐19, was demonstrated in the Netherlands, Germany, Luxembourg, Malta, Sweden, Finland, Spain and France.

Observed between‐list differences of the level of standardised amenable death rates in the EU countries are due to discrepancies in selected diseases and age limits. However, when assessing the effectiveness of health systems in examined countries, it has not changed significantly.

#### **4.3. The impact of AMIEHS and EUROSTAT's list on amenable mortality by cause of death in Slovakia**

The analysis examines whether age‐standardised rates of amenable mortality based on AMIEHS or EUROSTAT's list are significantly different by sex and causes of death in Slovakia. We apply both lists on data for 2014.

**Figure 1.** Amenable mortality across the European Union countries by the four lists of causes of death, 2014. *Source*: Own calculation based on the data from WHO mortality database.

In Slovakia, there was a considerable increase in the number of deaths considered amenable, from 9325 by the AMIEHS's list to 10,451 under the EUROSTAT's list. Of the additional 1126 deaths, 753 were for men and 373 for women. The increase occurred in all age groups, mostly after 55 years of age, and also not negligibly in the children aged from 0 to 4 years. The majority of the increase was due to the inclusion of respiratory diseases in the EUROSTAT's list that contributed 585 deaths of the 1126 deaths. The increase in the number of amenable deaths revealed that the total amenable mortality rates, as well as the rates for men and women, calculated by EUROSTAT's list were *significantly higher* (by 14.2% for men and 11.3% for women) than the rates under the AMIEHS's list. Generally, a difference is statistically significance if a confidence interval around rate non‐overlap with the interval around another (**Table 4**).

**Table 5** reflects the age‐standardised amenable mortality rates, based on the AMIEHS and EUROSTAT's list (with 95% confidence intervals) by broad cause group in Slovakia, 2014.

Besides the inclusion of respiratory diseases in the EUROSTAT's list, the increases in the number of deaths were also due to the inclusions of epilepsy contributing 96 deaths, diabetes with 37 deaths and misadventures to patients during surgical and medical care adding 15 deaths. Thus, Assessment of Avoidable Mortality Concepts in the European Union Countries, Their Benefits and Limitations http://dx.doi.org/10.5772/67818 85


*Note*: CI, confidence interval.

*Source*: Own calculation based on the data from WHO mortality database.

**Table 4.** Number of deaths and standardised amenable death rates based on AMIEHS or EUROSTAT's list in Slovakia, 2014.


*Note:* nc, non‐classified.

In Slovakia, there was a considerable increase in the number of deaths considered amenable, from 9325 by the AMIEHS's list to 10,451 under the EUROSTAT's list. Of the additional 1126 deaths, 753 were for men and 373 for women. The increase occurred in all age groups, mostly after 55 years of age, and also not negligibly in the children aged from 0 to 4 years. The majority of the increase was due to the inclusion of respiratory diseases in the EUROSTAT's list that contributed 585 deaths of the 1126 deaths. The increase in the number of amenable deaths revealed that the total amenable mortality rates, as well as the rates for men and women, calculated by EUROSTAT's list were *significantly higher* (by 14.2% for men and 11.3% for women) than the rates under the AMIEHS's list. Generally, a difference is statistically significance if a confidence interval around rate non‐overlap with the interval

**Figure 1.** Amenable mortality across the European Union countries by the four lists of causes of death, 2014. *Source*: Own

**Table 5** reflects the age‐standardised amenable mortality rates, based on the AMIEHS and EUROSTAT's list (with 95% confidence intervals) by broad cause group in Slovakia, 2014.

Besides the inclusion of respiratory diseases in the EUROSTAT's list, the increases in the number of deaths were also due to the inclusions of epilepsy contributing 96 deaths, diabetes with 37 deaths and misadventures to patients during surgical and medical care adding 15 deaths. Thus,

around another (**Table 4**).

84 Advances in Health Management

calculation based on the data from WHO mortality database.

Source: Own calculation based on the data from WHO mortality database.

**Table 5.** Standardised amenable mortality rates based on the AMIEHS and EUROSTAT's list (with 95% confidence intervals) by broad cause group in Slovakia, 2014.

additional causes of death included in the EUROSTAT's list accounted for 12.1%. A largest share in both lists is presented by ischaemic heart disease representing 44.9% under the AMIEHS's list and 40% in the EUROSTAT's list. However, standardised death rate of ischaemic heart disease has not changed when comparing the two lists. The other circulatory disease reported the statistically significant decrease of standardised death rates by 14.1% in the EUROSTAT's list contrary to the AMIEHS's list that was due to the exclusion of heart failure from the group. However, heart failure represented a substantial cause accounted for 14.1% in the group of other circulatory disease under the AMIEHS's list. In spite of the fact that infectious disease reflected the lowest numbers of deaths in the both lists, they recorded the largest statistically significant increase under the EUROSTAT's list because of the additional causes of death (tuberculosis, hepatitis C, selected invasive bacterial and protozoal infections) to the HIV contained in the AMIEHS's list. Moreover, in the HIV cause group, there was the extension of the age limit on the all age groups, whereas the age limit 0–74 years was included in the AMIEHS's list. In the neoplasms cause group, there was a statistically significant increase in the number of deaths by 15% mainly because of the addition of malignant neoplasms of skin and bladder cancer to the EUROSTAT's list and the shortness of the upper age limit of leukaemia. Finally, the standardised death rates for the surgical, congenital and perinatal conditions increased significantly under the EUROSTAT's list by 69.1 and 59%, respectively, mainly due to the inclusion of some surgical conditions (acute abdomen, appendicitis, intestinal obstruction, etc.) and the extension of the scope of congenital malformations to the overall 17 chapters of ICD‐10.
