This study was structured as 8 main stages including:

  1. Determination of the study area and local cases
  2. Data collection
  3. Database design and determination of annual incidence rates
  4. Classification of the incidence rates and mapping
  5. Interpretation of the thematic maps for determining the provinces where comprehensive examinations will be executed in the study area
  6. Statistical assessment of the incidence rates by geographic regions of Turkey, study area and high and low risk provinces
  7. Intensive statistical analysis for determining the temporal change of cases in provinces with high and low risk
  8. Determination of the correlation between water quality and quantity change with cases.

Although first seven stages were completed in the study, the last process was not be implemented by ITU since the required water quality/quantity data was not provided by the national governmental authorities although it was officially requested.

 

Determination of the study area and local cases        

Turkey with totally 1778 km of border to Black Sea is divided in to 81 administrative provinces in seven geographical regions, which are Marmara, Aegean, Black Sea, Mediterranean, Eastern Anatolia, Central Anatolia and South-Eastern Anatolia Regions (see Figure 2). Since the Black Sea Catchment was selected as the study area of the project, only 41 provinces of Turkey were completely or partially covered by the study area of EnviroGRIDS project. 25 provinces completely covered study area are Amasya, Artvin, Bilecik, Bolu, Çankırı, Çorum, Giresun, Gümüşhane, İstanbul, Kastamonu, Kırklareli, Kocaeli, Ordu, Rize, Sakarya, Samsun, Sinop, Tokat, Trabzon, Zonguldak, Bayburt, Bartın, Karabuk, and Düzce. Remaining 17 provinces partially covered by the study area are: Afyon, Ankara, Bursa, Erzincan, Erzurum, Eskişehir, Kayseri, Kırşehir, Konya, Kütahya, Nevşehir, Sivas, Uşak, Yozgat, Aksaray, Kırıkkale, and Ardahan. Provinces completely covered by the study area are mostly located in Black Sea and Marmara Regions along the Black Sea Cost of Turkey. Partially covered provinces are located in Marmara, Aegean, Central Anatolia and Eastern Anatolia Regions. The map presented in Figure 3 indicates the provinces partially or completely covered by the study area.

Figure 3. Geographical regions in Turkey

Figure 4. Coverage of study area in Turkey

Two of the Group A notifiable communicable water related diseases, Hepatitis A and Amebic Dysentery (A Dysenteriae), were considered as the cases on which detailed examinations will be performed, because these diseases has a stable notification system and notified data are more reliable than some other diseases considered in the same group. These diseases were also selected as the sample cases since they are commonly observed in Turkey as compared with the other diseases.

Data collection

Two communicable water borne diseases -Amebic (A) Dysentery (Amebiasis Dysenteriae) and Hepatitis A- were selected as main cases to be examined in this study. In this context, province based monthly recorded data covering the period from 2000 to 2009 were obtained from The Ministry of Health. Stated data includes total number of cases and deaths (by age and gender) of selected water borne diseases. In addition to health data, census data were obtained from The State Statistical Institute (SSI) for the years of 2000, 2007 – 2010. Gaps of census data for the period of 2001-2006 were filled by using geometrical increase method by using the formula presented in Equation 1. 

(1)

In addition to the health data, water quality and the quantity data, collected for the same period with health data in the provinces where specific examinations are executed, were officially requested from the related authorities such as The Ministry of Health, Local Health Authorities, and etc. stated data is planned to be used to introduce the relationship between water quality and quantity with health. However official request was refused by the Ministry of Health.

Geometric data used for mapping water borne communicable diseases is vector data covering administrative boundaries of the 81 provinces in Turkey. Vector data of the neighboring countries and main lakes in Turkey were also used as the complementary data for communicating the reference information about the location of Turkey in thematic maps. Universal Transversal Mercator (UTM) Coordinate System defined for zone 37 was used and World Geodetic System 1984 (WGS84) was selected as the geographic reference of the vector data. 

1.3 Database design and determination of annual incidence rates

In the third stage of the study collected data was organized.  In this context, a relational database designed by considering the 10 year health data (2000 - 2009) including case and deaths of water related diseases and census data and other attribute data. As it is presented in Figure 5 main table of the database was planned as the "Case_Death" table which includes monthly health data recorded at each province in Turkey. Stated table organizes the health data (Possible and absolute cases and total deaths by gender, age ranges, and illness id) and relations maintained with the tables containing the data of illnesses, cities (provinces) and etc.

 

Figure 5. Relational database

Conceptual database implemented by using SQL Server 2008 and collected attribute data imported in to SQL Server 2008 application (see Figure 6).

In epidemiology, progress of diseases are controlled by using relative data such as incidence rate rather than total number of cases for introducing the impact of the related disease over population. Incidence rate is defined as the number of new cases per population in a given time period (Hennekens et al., 1987). In this study annual incidence rates of diseases were calculated per 100 000 people for each province by dividing total number of the cases with population as indicated in following formula:

(2)

After data verification works were completed, annual incidence rates of each disease by province were calculated through database via SQL queries.

 

Figure 6. SQL Server 2008 application

1.4Classification of the incidence rates and mapping

Threshold values introducing endemic or epidemic characteristics of the disease are also important for monitoring the spread of the disease in epidemiology. These threshold values are usually numbers of cases in a defined period in excess of (a predetermined) expected number (WHO, 1999). They can be used for classifying the data to be mapped.

Incidence rates can also be used for setting threshold values instead of number of cases. The attainment of a threshold value should be considered as signaling an outbreak and should trigger specific responses in epidemiology. Determination of threshold values is considered as a case specific process therefore the specific threshold must be developed on the basis of local epidemiology and of immunization programme objectives. Two methods can be used for setting the threshold. Local epidemiology used to set up thresholds in the first method. In this context, a review of case reports from previous years (preferably at least 5 years excluding epidemic years) should be used to set up the threshold. The average number of cases or the average incidence rate, for a defined geographical area during a determined period of time in non-epidemic years can be taken as the threshold above which one should be alerted to the possibility of an outbreak (WHO, 1999). In the second method immunization programme objectives were used to set thresholds. First method was used to set threshold values in this study.  After determining the threshold values GIS technology was used to determine provinces with high and low risk of incidence rate.

Temporal trends of the selected diseases during 10 year time were examined for setting the threshold values by using two dimensional line charts. As it is indicated in Figure 7-10 each line in chart represents a province and X axis indicates time in terms of year from 2000 to 2009 while Y axis indicates annual incidence rates of Hepatitis A and A Dysentery per hundred thousand people. While examining the temporal trends, provinces of the study area were examined in two different classes as provinces completely covered by the study area and provinces partially covered by the study area for avoiding making mistakes while interpreting the complicated line graphs. Figure 7 and Figure 8 indicate the annual incidences of A Dysentery in provinces completely and partially covered by the study area respectively. Additionally, Figure 9 and Figure 10 indicate the annual incidences of Hepatitis A in provinces completely and partially covered by the study area respectively.

 

Figure 7. Annual incidences of A Dysentery in provinces completely covered by the study area

 

Figure 8. Annual incidences of A Dysentery in provinces partially covered by the study area

Figure 9. Annual incidences of Hepatitis A in provinces completely covered by the study area

 

Figure 10. Annual incidences of Hepatitis A in provinces partially covered by the study area

When the line graphs are interpreted, diseases were sighted in endemic level in some provinces, while they occur with high incidence rates in some other provinces. For example while hepatitis A cases occur with high incidence rate in Bayburt and Ordu, it occurs endemically in İstanbul and Artvin (see Figure 9).

Line graphs were also used to determine the threshold values for endemic, hyper endemic and epidemic occurrence levels for both of the selected diseases. As it is sampled in Figure 11 for A Dysentery, stated thresholds were set by considering the temporal trend of the disease in each province of the study area as explained by WHO (1999).  In this context, a review of case reports, as annual incidence rates, from previous years (115 months approximately 10 year period) were used to set up the threshold. The stable incidence rate for study area was considered as the threshold between endemic and hyper endemic levels. The incidence rate which is rarely exceeded in the study area was considered as the threshold between hyper endemic and epidemic levels. As it is indicated in Table 1, endemic level was considered as the level with low risk of incidence rate originated by the occurrence of the disease, while hyper endemic and epidemic levels have high risks of incidence rate.

Table 1. Threshold values

Illness Name

Thresholds of Incidences (%000)

Endemic

Hyper Endemic

Epidemic

Low Risk

High Risk

A Dysentery

≤ 10

> 10   ˄    ≤ 20

> 20

Hepatitis A

≤ 10

> 10   ˄    ≤ 20

> 20

 

Figure 11. Endemic, hyper endemic and epidemic levels for A Dysentery

As the following step of the study GIS technology were used to produce thematic maps for examining the temporal change of annual incidence rates of diseases considered in this study. In this context, the vector data was imported into a GIS environment together with the tabular data including annual incidences of diseases and links between these data were provided. Data classification tools were used to classify and visualize the data geographically depending on determined threshold values indicated in Table 1. Each class was visualized by using value of the color and considered values (tones of red color) were represented in Table 1 for endemic, hyper endemic and epidemic levels. Produced thematic maps indicating the annual incidence rates of hepatitis A and A Dysentery are presented in Figure 12 and Figure 13 respectively.

1.5Interpretation of maps

Designed thematic maps indicating the annual incidence rates of Hepatitis A and A Dysentery were interpreted for determining the provinces where comprehensive examinations will be executed in the study area. These provinces were considered as provinces with high and low risk of incidence rate: totally 16 provinces were determined (8 as provinces with high risk and 8 as provinces with low risk of disease occurrence) by considering the temporal trends of annual incidence rates of selected diseases in the provinces. Provinces which have a stable trend of hyper endemic and epidemic occurrences of diseases in both selected diseases were considered as high risk provinces while the provinces with a continuous trend of endemic occurrence were determined as low risk provinces. As a result, Kastamonu, Nevşehir, Kütahya, Ordu, Bayburt, Ordu, Ardahan, Rize, and Kırklareli determined as provinces with high risk while İstanbul, Bilecik, Kayseri, Giresun, Tokat, Afyon, Eskişehir, and Artvin considered as provinces with low risk.

1.6Statistical assessment of the incidence rates

Incidence rates were statistically assessed by geographic regions, provinces in or out of the study area and high and low risk provinces respectively. In this context:

  • one-way ANOVA (Analysis of Variance) were used for comparing mean incidence rates between more than two different data sets,
  • Tukey test was used as the Post Hoc Test for determining the origin of the possible differences introduced in one-way ANOVA,
  •  Student t test (two independent samples test) was used to compare mean incidence rates.
  • Spearman correlation analysis was used to determine temporal change of incidence rates (relation between year and incidence rate).
  • Sign test was used for examining the temporal change in case numbers of each disease in province level.

Figure 12. Temporal changes in incidence rates of hepatitis A

Figure 13. Temporal changes in incidence rates of A Dysentery