Health Outcome Data Review
TRI COUNTY LANDFILL
PINE AND LIBERTY TOWNSHIP, MERCER COUNTY, PENNSYLVANIA
The Pennsylvania Department of Health (PADOH) prepared this Health Consultation (HC), under a Cooperative Agreement with the Agency for Toxic Substances and Disease Registry (ATSDR), to address community concerns of health risks associated with the Tri-County Landfill. Area residents had expressed concerns of elevated cancers, infant deaths and miscarriages that they believed were associated with the landfill. In response to these concerns, the PADOH evaluated available health outcome data and has determined that there are no excess health problems in the study area from what would be expected when compared with Mercer County or the state of Pennsylvania as a whole.
The Tri-County Landfill is located at Pine and Liberty Townships, Mercer County (Figure 1, 2). This Health Consultation (HC) is one of two HCs that PADOH is preparing for the site. This HC focuses on the health outcome data analysis in response to the community's health concerns of cancer mortality, cancer incidence, infant death, fetal death and miscarriages.
Without sufficient evidence of exposure population, level of exposure, or exposure pathways, we simply selected Pine and Liberty Townships where the landfill is located as the study area for this health outcome data analysis. Analysis of data at this level is used as a screening tool for identifying unusual health problems that occur more often than is expected compared with Mercer County or the State of Pennsylvania.
Sources of Data 
This section discusses the types of health data we used for the analysis, which include: cancer death, cancer incidence, infant death, fetal death and live birth data.
Pennsylvania's Certificate of Death is the source document for the cancer mortality and infant mortality data used in this HC. Infant mortality was defined as death of an infant less than one year of age. Death certificates are usually completed by hospital personnel, physicians and funeral directors. Mortality data included all those individuals at the time of their deaths who had recorded homes within Pennsylvania.
The Pennsylvania Cancer Registry (PCR) Report Form is the source document for the cancer incidence data used in this HC. A cancer incidence is each newly diagnosed case of cancer in Pennsylvania. Hospital personnel report the primary cancer "site" for each incidence by completing the PCR form forwarded to the registry. A primary "site" is the primary organ or tissue of the body where the cancer is believed to have originated. A cancer patient may have more than one primary site. In this HC, the number of cancer incidences refers to the number of primary sites reported, not the number of people, though the number of people and the number of primary sites are believed to be very close. The PCR captures all cancer incidence data with the exception of some specific types of skin cancer.
Pennsylvania's Certificate of Fetal Death is the source document for fetal death data used in this HC. Pennsylvania law defines fetal death as a product of conception, which is expelled or extracted from its mother after 16 weeks gestation.
Pennsylvania's Certificate of Birth is the source document for live birth data used in this HC. Pennsylvania law defines live birth as a product of conception , which is expelled or extracted from its mother which shows any evidence of life at any moment after such expulsion or extraction.
In this HC, each cancer occurrence in the study area was analyzed by the type of cancer death or the primary cancer site for cancer incidence. Twenty three (23) primary types of cancer were analyzed [Appendix I]. Indirect Method and Direct Method of standardization were used for analysis. Standarized Mortality Ratio (SMR) and Standarized Incidence Ratio (SIR) were calculated for comparison. (See explanation of methods in Appendix II.)
Multiple-year data are generally used in analysis because cases collected in a single year are subject to a large amount of variation from year to year due to chance. In this HC, ten (10) years of cancer mortality (1986-1995) and cancer incidence (1985-1994) data, and seventeen (17) years of birth, infant death and fetal death (1980-1996) data were evaluated.
Using the Indirect Method, numbers of observed and expected cancer deaths for each type of cancer from 1986-1995 are compared in Table 1. Among the 5,121 residents in the Pine and Liberty Townships (53.5% male, 46.5% female), overall cancer mortality is similar to what is expected (113 observed vs. 102 expected, SMR=1.11). However, when reviewing each individual types of cancer, Non-Hodgkin's Lymphoma (NHL) was the only type of cancer that showed an increase beyond statistical expectation ( 12 observed vs. 4.28 expected, SMR=2.80).
Using the Direct Method, cancer mortality rates for the time period of 1990-1994 are compared for the study area, Mercer County, and the State of Pennsylvania as presented in Table 3. None of the rates for the study area show any unusual or unexpected elevations when compared to the rates of Mercer County and the State of Pennsylvania as a whole. NHL mortality did not show a higher rate for this five-year comparison time period.
Numbers of observed and expected cancer incidences are compared in Table 2 using the Indirect Method. A total of 126 new cancer cases were diagnosed over the ten year time period. The number of new cancer cases for both male and female were much lower than expected. The expected number of cancer incidences among men was 121.69 compared to 59 observed with a SIR of 0.48. The expected number of cancer incidences among women was 104.99 compared to 67 observed with a SIR of 0.65.
Infant deaths and Miscarriages
Local residents expressed concerns of approximately six miscarriages and five infant deaths mostly due to congenital anomalies experienced in the past ten years.
From 1980-1995, a total of 942 live births and nine infant deaths were reported to the PADOH in the Pine and Liberty Townships. Thus, the annual average of infant deaths for the 16-year period is 9.6 per 1,000 live births ( ±6.3 C.I.) ranging from 3.3 to 15.9. For the same time period, Pennsylvania reported the annual average rate ranging from 7.7 to 13.2 per 1,000 live births in the state as a whole. The occurrences of infants deaths in the townships do not appear to be elevated when compared to the State of Pennsylvania.
In Pennsylvania, non-induced fetal deaths under 16 weeks gestation are not required to be reported. That is, miscarriages that occurred in the early trimesters were not known. So no comparisons or conclusions about the rates of miscarriage or fetal death in the early trimesters can be made.
For estimating miscarriages that occurred after 16 weeks gestation, the fetal death data were used for analysis. From 1980-1995, a total of seven non-induced fetal deaths (over 16 weeks gestation) were reported in the Townships. The calculated annual average rate of fetal death in the Townships was 7.4 per 1,000 live births ( ±5.5 C.I.) ranging from 1.9 to 12.9. In comparison, the annual rate of fetal deaths ranged from 15.4 to 21.3 per 1,000 live births in Pennsylvania for the same time period. The occurrences of miscarriages in the Townships do not seem to be elevated compared to the State of Pennsylvania as a whole.
Cancer Mortality and Cancer Incidence
This review was based on a population of 5,121 residents living in Pine and Liberty Townships. Comparisons of observed and expected number of cancer deaths and cancer incidences were also carried out for each primary type of cancer. In no instance was the difference between the observed and expected numbers of cancer deaths or incidences in the study area more than expected except for the Non-Hodgkin's Lymphoma (NHL) deaths during the study period. NHL mortality was the only type of cancer that showed an increase beyond statistical expectation. Although NHL deaths were found to be elevated in the study areas, the NHL incidence was not found to be more than expected.
In comparison, since the early 1970's, the incidence of NHL has been increasing more rapidly than any malignancy except malignant melanoma both in men and women in the nation [3,4]. However, this increase cannot yet be adequately explained. Some of the increased incidences in NHL may be explained by improvements in medical diagnosis over the past 50 years. A study found that in humans, certain viral infections have been associated with NHL such as the AIDS virus . The same study also found that drugs that dramatically suppress the immune system increase the risk of developing NHL (e.g. drugs given to organ transplant or cancer patients). The risk for NHL is also increased by a family history of certain other cancers [6, 7]. When reviewing the literature regarding NHL and pesticide exposures in 1992, Zahm and Blair noted a modest increase in risk for agricultural workers in general, but a much higher risk for those workers exposed to specific pesticides . However, no pesticides were detected at the Tri- County Landfill site. Other studies also indicated that the use of hair dyes increases the risk for NHL [9, 10].
Although all the above factors combined still cannot explain the entire national increase in NHL incidences, the largest increase was observed among persons aged 65 and older . This observation is also valid in the study area where all those who died of NHL were above the age of 65. Due to no further information available from the Death Certificate on individual's risk factors, other exposures, if any, or residential history, the causes of the elevated NHL mortality in the Townships are not known. Based on the characteristics of this disease and its known causes, they are not believed to be site related.
We often hear comments that more cancers are seen now than decades ago. First of all, with a few exceptions, primarily lung and AIDS-related cancers, there has been little overall increase in the number of new cases of cancer reported or the number of cancer deaths over the last 40 years in the nation . Also, modern screening methods, such as mammography for breast cancer and the prostate-specific antigen (PSA) test for prostate cancer, create the appearance of a sudden increase in new cancer cases. There is no correspondingly large increase in mortality from these forms of cancer, indicating that we are finding more previously undetected malignancies . Second, cancer is not a rare disease nor a single disease, but a large group of over 100 diseases . The life time probability of developing cancer is estimated at one in three. Different types of cancer have unique characteristics and are associated with different risk factors. Therefore, it is important to evaluate individual cancer cases by their cancer type. Third, such factors as diet, sunlight, genetic history, smoking and other environmental exposures can all play a part in causing cancer. Fourth, anyone can get cancer including an unborn baby to adults well over 90 years old. Cancer has been called the disease of the aged because cancers develop more commonly in the elderly. Cancers usually take many years to develop, sometimes 25 or more, which is one reason they are found more often in the elderly. The American population now includes a higher portion of elderly persons than ever before. Finally, many of the cancer cases can be prevented through early detection and prevention measures.
Lung cancer is the leading cause of cancer death in the United States . This is also true for the Pine and Liberty Townships. Trachea, bronchus, lung, and pleura cancers were identified as the leading causes of cancer death with a total of 80 from 1986-1995, though not more than the statistical expectation (71.77 expected). Accounting for 31% of cancer deaths, trachea, bronchus, lung, and pleura cancer should be and can be reduced or prevented. Smoking cessation, radon reduction, and prevention have proven tremendous potential to reduce the mortality associated with respiratory cancer .
Colon cancer is the second leading cause of cancer death in men and the third leading cause of cancer death in women nationally and in the study area. Increased risk of colon cancer has consistently been associated with a diet high in saturated fat (especially through meat intake), low in vegetables and low in high-fiber grains. Growing evidence also suggests that a lack of either occupational or recreational physical activity increases the risk of colon cancer. To reduce colon cancer, it was recommended to adopt a diet of five or more servings of fruits and vegetables daily, five or more servings of whole grains and/or beans daily, and a selection of the lowest-fat choices within each food group and to increase physical activities .
Breast cancer is the second leading cause of cancer death in women in the study area. Preventive measures are crucial to reduce breast cancer. Studies have demonstrated that clinical breast examination by a physician or nurse and mammography screening are effective methods for early detection of breast cancer. It was estimated about one of every nine women will develop breast cancer at some time during her lifetime while failure to use mammography on a population basis is believed to account for as much as 19% to 25% of all breast cancer deaths .
Infant and Fetal Mortality
Even though the statistics did not suggest that infant or fetal deaths due to birth defects are elevated in the Pine and Liberty Townships, families experienced great emotional stress from their loss. The Centers for Disease Control and Prevention (CDC) has recently developed certain national guidelines regarding preventing common birth defects that affect thousands of babies each year. Each year about 4,000 pregnancies in the U.S. are affected by birth defects of the spine and brain; namely spina bifida and anencephaly. CDC recommends that all women of childbearing age should consume 0.4 mg of folic acid daily . CDC estimates that up to 3,000 of these birth defects could be prevented every year if women consumed folic acid before and early in pregnancy.
Like any statistical review, there are certain limitations as to its capabilities. First, the quality of data presented in this HC is directly related to the completeness and accuracy of the information reported. PADOH utilizes quality assurance measures to ensure that the data are collected and coded as completely and accurately as possible.
Second, this HC intends to determine if there is an increased rate of illness. Cause and effect relationship between the site and health effects cannot be established by this HC because we are not able to exclude confounding factors. They are the other important causal factors which can considerably contribute to cancers and other adverse health effects. For example, "lifestyle" factors such as smoking, high fat and low fiber diet, alcohol consumption, and lack of physical activity have been shown to be the major risk factors for the majority of the cancer cases and other health conditions. Heredity, or family history, is also an important factor. Chemical exposures from various sources, especially occupational exposure, in combination with other factors, may contribute to various cancers and adverse health effects.
Third, the data source documents recorded the residence in which the individual resided at the time of death or diagnosis. Information on migration, occupation, previous residence or length of residency, and other risk factors is not contained in these documents. Thus, it is not possible to determine individual's residential history, past environmental exposure, if any, and other risk factors which can be contributing factors to the health effects area residents may experience later on.
Finally, rates based on an unusually small number of events over a specified period of time or sparsely populated geographic area, can result in unstable rates computation. Some types of cancer were very small (<5 observed cases) and made statistical estimation rather unstable and difficult to interpret. One way to add more credence to a computed rate is to expand the period of time to enlarge the absolute numbers. The width of the confidence interval decreases as the sample size (number of observed cases) increases which provide more accurate estimation of the rates. Multiple years of Township data, 20 years or more, will be ideal for rate computation. However, for the purpose of comparability, the period of study was conducted on a 5-year basis.
On December 9, 1997, PADOH staff met with local residents to gather information and listen to their concerns. The following health concerns were presented at that meeting or through telephone or written communications:
- Concern: Local residents expressed concerns that the landfill is responsible for the adverse health outcomes that people living near the landfill have experienced. The discussions included concerns regarding the number of congenital birth defects, miscarriages, and cancers experienced by neighbors of the landfill. An extensive list of residents, addresses, and health problems was provided to the Health Assessment Program staff.
- Concern: There was a health study completed on children residing in the trailer park off Amsterdam Road in 1985-1986 because of the number of childhood illnesses at the time.
Response: Health related concerns are addressed in this HC. PADOH evaluated the cancer mortality, cancer incidence, infant death, fetal death and live birth data and determined that there are no excess health problems in the area from what would be expected when compared to Mercer County or the state of Pennsylvania as a whole. NHL mortality is the only type of cancer found elevated. We do not know the causes for the elevation but believe they are not related to the site.
Response: The Health Assessment Program is interested in reviewing the study. A local resident will forward additional information on the study so that a copy can be obtained for our review.
The review of the available health outcome data in this HC did not indicate an excess of cancer deaths, cancer incidences, infant death or miscarriages except NHL mortality for the time period of 1986-1995. The causes for the elevation are not known, however, they are not believed to be associated with the site. In conclusion, the authors did not observe a clustering of health problems related to the site.
- Vital Statistics Data Files. State Center of Health Statistics and Research, Pennsylvania Department of Health.
- Cancer Epidemiology and Prevention. David Schottenfeld, New York, 1996.
- Chronic Disease Epidemiology and Control. American Public Health Association, 1993.
- Devesa SS, Fears T. Non-Hodgkin's Lymphoma time trend: United States and International Data. Cancer Research (Suppl.) 1992; 52 (19):5432s-5440s.
- Hartge P, Devesa SS. Quantification of the impact of known risk factors on time trends in non-Hodgkin's lymphoma incidence. Cancer Research (Suppl.) 1992; 52 (19): 5566s-5569s.
- Pottern LM, Linet M, Blair A, Dick F, Burmeister LF, RG, Schuman LM, Fraumeni Jf. Familial cancers associated with subtypes of leukemia and non-Hodgkin's Lymphoma. Leukemia Research 1991; 15 (5):305-314.
- Linet MS, Pottern LM. Familial aggregation of hematopoietic malignancies and risk of non-Hodgkin's Lymphoma. Cancer Research (Suppl.) 1992; 52: 5468s-5473s.
- Zahm SH, Blair A. Pesticides and non-Hodgkin's lymphoma. Cancer Research (Suppl.) 1992; 52:5485s-5488s.
- Pearce N, Bethwaite P. Increasing incidence of non-Hodgkin's Lymphoma: Occupational and environmental factors. Cancer Research (Suppl.) 1992; 52:5496s-5500s.
- Zahm SH, Weisenburge DD, Babbitt PA, Saal RC, Vaught JB, Blair A. Use of hair coloring products and the risk of lymphoma, multiple myeloma, and chronic lymphocytic leukemia. American Journal of Public Health 1992; 82:990-997.
- MMWR, Centers for Disease Control and Prevention, February 27, 1998.
- Fisher A, Worth W, Mayer D. Update: Is there a Cancer Epidemic in the United States? American council on Science and Health.
Alice Hoffman, M.H.A.
Division of Environmental Health
Pennsylvania Department of Health
This Health Consultation for the Tri-County Landfill site was prepared by the Pennsylvania Department of Health under a cooperative agreement with the federal Agency for Toxic Substances and Disease Registry (ATSDR). It is in accordance with approved methodology and procedures existing at the time the health consultation was initiated.
Technical Project Officer, SPS, RPB, DHAC
The Division of Health Assessment and Consultation (DHAC), ATSDR, has reviewed this health consultation and concurs with its findings
Chief, SPS, SSAB, DHAC
Tables 1 - 3 were not available in electronic format for conversion to HTML at the time of preparation of this document. To obtain a hard copy of the document, please contact:
Agency for Toxic Substances and Disease Registry
Division of Health Assessment and Consultation
Attn: Chief, Program Evaluation, Records, and Information Services Branch E-56
1600 Clifton Road NE, Atlanta, Georgia 30333
COMPARABILITY TABLE OF ICD-O-2 AND ICD-9 CODES for 23 PRIMARY CANCER SITES
|Buccal Cavity and Pharynx||C00.0-C14.8||140.0-149.9|
|Rectum, Anus, Rectosigmoid||C199.9, C20.9, C21.0-C21.8||154.0-154.8|
|Trachea, Bronchus, Lung, Pleura||C33.9, C34.0-C34.9, C38.4||162.0-163.9|
|Melanoma of Skin||C44.0-C44.9 and
M-8720 to M-8780
|Female Breast||C50.0-C50.9 *||174.0-174.9|
|Cervix Uteri||C53.0-C53.9 **||180.0-180.9|
|Kidney and Renal Pelvis||C64.9, C65.9||189.0-189.1|
|Brain and Other Nervous System||C70.0-C72.9||191.0-192.9|
|Non-Hodgkin's Lymphomas||M-9590 to M-9595,
M-9670 to M-9687,
M-9690 to M-9709,
M-9711 to M-9714,
M-9740 to M-9741
|159.1, 200.0-200.8, 202.0-202.2, 202.6, 202.8-202.9|
|Hodgkin's Disease||M-9650 to M-9667||201.0-201.9|
|Multiple Myeloma||M-9731 to M-9732||203.0, 203.8|
|Leukemias||M-9800 to M-9941|
** Excludes in situ cases.
Note: Sites Buccal Cavity to Thyroid, as listed above, exclude sites M9590 to M9980 of the ICD-O (FT) and M9590 to M9989 of the ICD-O-2.
In order to know whether the townships had high cancer mortality, the observed number of cancer deaths for that area must be compared with something else. In this case, the State of Pennsylvania was used. The cancer mortality rate for the state as a whole was used to calculate an expected number of cancer deaths that would have occurred in the townships over the same period of time. This "expected number" should not vary significantly from the townships' cancer death numbers if the township rates are similar to Pennsylvania.
The average annual Pennsylvania age-sex-specific cancer mortality rate for selected cancers from 1988-1992 was applied to the township population to obtain the expected numbers. The observed and expected numbers of cancer deaths are compared to each other by computing a ratio of observed to expected number. This ratio is the Standardized Mortality Ratio (SMR). The results of the analysis are presented in Table 1.
When the SMR is less than 1.0, fewer cancer deaths have been observed than expected. When the SMR is greater than 1.0, there are more cancer deaths than were expected. The 95% confidence interval is the statistical test used to evaluated whether the difference between the observed and expected numbers may be due to random variation. If the confidence interval does not include 1, then the difference between the observed and expected is considered to be "statistically significant" (Bailar and Ederer, 1964).
Likewise, the evaluation of cancer incidence was performed using Standardized Incidence Ratios (SIR), that is, the ratio of the observed number of cancer incidence divided by the expected number. The results are presented in Table 2.
Caution should be exercised, however, when interpreting a SMR or SIR. The interpretation must take into account the actual number of cases observed and expected, not just the ratio. Two SMRs or SIRs can have the same values but not the same stability. For example, an SMR of 1.5 could mean 3 cases were observed and 2 were expected (3/2=1.5). Or it could mean 300 cases were observed and 200 were expected (300/200=1.5). In the first instance, only one excess cancer occurred, which could have easily been due to chance. But, in the second instance, 100 excess cancers occurred and it would be less likely that this would occur by chance alone. This shows that rates are "less stable" if the number of cases is small.
Area residents requested PADOH to conduct a cancer rate comparison of the townships with Mercer County and the State of Pennsylvania. For comparability, age-standardized rates were used to compare cancer mortality between the townships surrounding the site, Mercer County and the state of Pennsylvania. The direct Method of Standardization was utilized.
Using this method, age-specific mortality rates for the Township (study area) and the County and the State (comparison areas) were applied to a standard population in order to calculate what rates would be expected if both populations had the same age distribution as the standard population. The expected events are totaled and divided by the total of the standard population and multiplied by 100,000 yields the age-adjusted rate per 100,000 population. The standard population used in the calculation of age-standardized rates is the 1970 U.S. standard million population. This is the standard population used, too, in the Pennsylvania Annual Report of Cancer Incidence and Mortality published by the State Center of Health Statistics. The annual report also utilizes data of five year blocks to report the rates for the counties and the state. Therefore, we selected the rates of the same time period 1990-1994 for the township to allow comparability. The results are presented in Table 3, 4.
By using age-standardized rates, we can be sure that any differences in the cancer rates of the two areas being compared are not due to differences in the age compositions of their respective populations. Since an age-standardized rate is a statistical measurement, it cannot be compared to any other type of rate or be considered an absolute measurement of events of interest.
For any statistical measurement, the reported number of events (cancer deaths or cancer incidences) should be viewed as an estimate of the "true" number in the population, subject to chance variation. Thus, an observed rate (calculated by using reported number of events ) would not be considered a "true" rate, but a point estimate. To test whether the difference between rates is statistically significant, 95% confidence intervals were calculated for each point estimate. Differences between the comparison rates would be considered significant at the 95% confidence level if the confidence intervals of the point estimates do not overlap.