# Appendix D: Health Outcome Data Review

**Methods**

The review of health outcome data in reference to Portsmouth and Newington, in addition to an individual’s health concern in New Castle, New Hampshire, has encompassed an analysis of available cancer information from the New Hampshire State Cancer Registry (NHSCR). The NHSCR is located at Dartmouth Hitchcock Medical Center in Lebanon, New Hampshire. The NHSCR maintains statistics regarding 23 types of cancer in New Hampshire. The NHSCR is in the process of updating their records through 1994.

The cancer information for Portsmouth (including Newington) and New Castle, NH was requested for the years that cancer incidence rates were available (1987 through 1991) from the Bureau of Vital Records and Statistics, which maintains a close working relationship with the NHSCR. Once the data were received, a descriptive epidemiological analysis was conducted. The data were reviewed according to the following factors: sex and age (five-year age groupings; ages 0-85+) for each type of cancer reported in Portsmouth, NH (1990 census population, 25,925 persons) and New Castle, NH (1990 census population, 840 persons) between 1987 and 1991. The types of cancer reported during that time period in Portsmouth, NH are the following: cancer of the esophagus, colon, rectum, lung, melanoma, breast, cervix, uterine, ovary, prostate, testis, bladder, kidney, thyroid, non-Hodgkin’s lymphoma, stomach, buccal cavity, and pharyngeal; and, in New Castle, NH: female breast was the only cancer reported over the same time period. Only four cancer (kidney, melanoma, leukemia, and cervical cancer) patients reported a residential address at Pease Air Force Base between 1987 and 1991. The small number of cancers reported at Pease Air Force Base, as well as the fact that the base population was predominantly transient prohibited further analysis.

Very few cancer cases were observed in Newington, NH (1990 census population, 990 persons) between 1987 and 1991, thus preventing further analysis. A series of calculations enabled us to review the observed number of specific cancer cases in the areas of study and compare that value to the expected number of cancer cases. This exercise compared the number of cancer cases among males and females in a variety of age groups in both Portsmouth, NH and New Castle, NH with the state of New Hampshire as a whole.

It is important to note that the quality of the NHSCR cancer data is heavily relied upon to conduct a thorough descriptive epidemiological analysis. Unfortunately, we were unable to examine the recent trend of cancer incidence in Portsmouth, Newington, and New Castle, NH due to the unavailability of more current cancer statistics from the NHSCR (i.e., 1992-present).

**Results**

All rates are within the expected range at the 95% confidence interval in Portsmouth, NH. In general, the data reviewed do not demonstrate a statistically significant elevation of cancer incidence within this community. However, two cancers were shown to be elevated in Portsmouth, NH between 1987 and 1991: cervical cancer in females and non-Hodgkin’s lymphoma in males. Although only 7.7 cervical cancer cases were expected in Portsmouth, NH during that time interval, 25.0 cervical cancer cases were observed. There were also 4.8 non-Hodgkin’s lymphoma cases in males expected in Portsmouth, NH during that time interval, but 11.0 non-Hodgkin’s lymphoma cases observed.

In addition, all rates are within the expected range at the 95% confidence interval in New Castle, NH. Overall, the data reviewed do not demonstrate a statistically significant elevation of cancer incidence within this community. None of the reportable cancers in New Castle, NH were statistically elevated between 1987and 1991.

**Discussion**

*Portsmouth, NH: Cervical Cancer*

According to the American Cancer Society, an estimated 13,700 cases of invasive cervical cancer were diagnosed in 1998 throughout the United States. Rates of cervical cancer incidence have steadily decreased over the past several decades, declining from 14.2 per 100,000 in 1973 to 7.8 per 100,000 in 1994. In addition, an estimated 4,900 cervical cancer deaths will occur in the United States in 1998. Death rates from cervical cancer declined 45% between 1972-1974 and 1992-1994 (American Cancer Society 1996).

Risk factors for cervical cancer are closely linked to sexual behavior and to sexually transmitted infections with certain types of human papillomavirus. Women who have first intercourse at an early age, multiple sexual partners, or partners who have had multiple sexual partners are at increased risk for developing the disease. Additional risk factors include cigarette smoking and low socioeconomic status (American Cancer Society 1996).

The American Cancer Society recommends that all women have Pap tests done by health care professionals as part of a routine pelvic exam. This test should be performed annually with a pelvic exam in women who are, or have been, sexually active, or who have reached the age of 18 (American Cancer Society 1998). New Hampshire Breast and Cervical Cancer Early Detection Program offers educational programs on breast and cervical cancer and is trying to improve access to medical care for women of all socioeconomic classes throughout the State of New Hampshire.

*Portsmouth, NH: Non-Hodgkin’s Lymphoma*

According to the American Cancer Society, an estimated 62,500 new cases of lymphoma will occur in the United States in 1998, including 7,100 cases of Hodgkin’s disease and 55,400 cases of non-Hodgkin’s lymphoma. Within the last 25 years, the incidence rates for non-Hodgkin’s lymphoma have nearly doubled. In addition, the American Cancer Society estimates that 24,900 deaths in the United States will be attributed to non-Hodgkin’s lymphoma (American Cancer Society 1996).

The risk factors for lymphoma are relatively unknown, but are believed to involve reduced immune function and exposure to certain infectious agents. People with organ transplants are at higher risk of developing lymphomas due to altered immune function. In addition, human immunodeficiency virus (HIV) and human T-cell leukemia/lymphoma virus-I (HTLV-I) are associated with increased risk of non-Hodgkin’s lymphoma. Other possible risk factors for lymphoma include occupational exposures to herbicides and solvents. Also, the antiseizure drug Dilantin may cause noncancerous overgrowths of lymphoid tissue, but these growths usually regress if the drug is discontinued. Some patients treated with this drug may develop non-Hodgkin’s lymphoma, but the risk is very small. (American Cancer Society 1996).

In addition, chemotherapy may increase a patient’s risk of developing leukemia or non-Hodgkin’s lymphoma 5 to 10 years following treatment. This is an important point for patients who have been treated for Hodgkin’s disease, because their risk of developing non-Hodgkin’s lymphoma is 4%-5% over a 10 period. Patients treated with radiation therapy for some other cancers have a slight risk of developing non-Hodgkin’s lymphoma later in life. However, it usually takes several years for this to happen, so these secondary cases of non-Hodgkin’s lymphoma are more common in adults than in children. Patients treated with both chemotherapy and radiation treatment are more likely to develop secondary leukemias or non-Hodgkin’s lymphomas (American Cancer Society 1996).

Furthermore, patients with transplanted organs, such as kidney, heart, and liver, are treated with drugs that compromise their immune system to prevent their immune system from rejecting the transplanted organs. This intentional immune suppression carries a significant risk for the patient of developing non-Hodgkin’s lymphoma. The exact risk of developing this lymphoma is dependent upon the type of drugs used (American Cancer Society 1996).

*It is important to keep in mind that most adults and children diagnosed with non-Hodgkin’s lymphoma have no well-defined, known risk factors. These parents or their children could have done nothing to prevent this type of lymphoma* (American Cancer Society 1996).

**Conclusions**

Two types of cancer that an elevated incidence rate in Portsmouth, NH between 1987-1991: cervical cancer in females and non-Hodgkin’s lymphoma in males.

First, no known risk factors for cervical cancer are associated with exposure to chemical contaminants in the environment. The known risk factors associated with cervical cancer, as previously mentioned, include the following: early age at first intercourse, multiple sex partners, human papillomavirus, and cigarette smoking. Pease Air Force Base is not likely the cause of the observed cervical cancer incidence in Portsmouth, NH.

Second, no well-defined, known environmental risk factors are associated with the development of non-Hodgkin’s lymphoma. As previously mentioned, possible environmental risk factors for lymphoma include occupational exposures to herbicides and other chemicals such as volatile organic solvents. A causal relationship between these risk factors and the presence of non-Hodgkin’s lymphoma does not currently exist. The American Cancer Society states that parents or their children can do nothing to prevent this type of cancer (American Cancer Society, 1996).

It is important to note that elevated rates of a particular illness in a community may not necessarily be attributed to site contamination, nor does it establish a link or imply causality with environmental contamination identified at a site. Many factors influence the development of disease, including personal lifestyle, occupation, and socioeconomic status. Previously in this document, pathways of human exposure to chemical contaminants at Pease AFB were evaluated and determined to be unlikely to result in adverse health effects. As a result, the increased incidence of non-Hodgkin’s Lymphoma observed in the Portsmouth area is not associated with environmental contamination originating from Pease Air Force Base.

**Recommendations**

Since regular Pap test screening is effective in reducing cervical cancer mortality, we support the American Cancer Society’s recommendation that women who are 18 years of age, or who are or have been sexually active have annual Pap smear tests. Two facilities in Rockingham County that offer such medical services that are based on the woman’s ability to pay: Women’s Health Consortium (603-431-1669) and Planned Parenthood of Northern New England (603-352-6898). In addition, the Breast and Cervical Cancer Early Detection Program located in the Department of Health and Human Services may also provide further information regarding additional resources, education, and outreach programs (1-800-852-3345 extension 4931).

*Evaluating Cancer Data and Basic Cancer Rate Terminology*

Cancer incidence is the number of new cases of cancer, by specific type, that is reported for a particular area over a specific period of time. A review of the cancer incidence for selected cancers can help determine whether the community is experiencing greater than normal levels of cancer.

The American Cancer Society estimates that approximately 8 million Americans alive today have a history of cancer. For men, the lifetime risk of developing cancer is 1 in 2; for women, the lifetime risk is 1 in 3. In 1998, about 1,228,600 new cases of cancer are expected to be diagnosed.

To help determine whether a community is experiencing a greater than expected rate of cancer, cancer statistics can be reviewed. First, it is necessary to verify the number of cases of cancer that actually occurred in the community. This is referred to as the observed cases. Observed cases are found through a cancer registry, in this case, the New Hampshire State Cancer Registry. Next, it is necessary to calculate the expected cases, which are the number of cases that would be anticipated to occur. The expected cases are a mathematical prediction of the number of cases that would be expected in a particular community population based on the number of cases that have occurred in a reference population, such as a metropolitan area, a state, or the nation as a whole. Prediction of the expected cases takes into account the age, sex, and race of persons in the community and assumes that the community population is similar enough to the reference population that the same proportion of any given cancer will be reflected in the community population. In the case of those living in Portsmouth, NH and New Castle, NH, the population of the state of New Hampshire is used as the reference population.

If the reference population has 1000 persons and 25 cases of cancer, the proportion of cancer in that population would be 25 in 1000 (25/1000) or .025. If the community has a population of 100, the method to calculate the expected number of cancers in the community would be to multiply the population of the community, 100 by .025, the proportion seen in the reference population (100 x .025). The expected number of cancers in the community would be 2.5.

If the reference population has 1000 persons and 25 cases of cancer, the proportion of cancer in that population would be 25 in 1000 (25/1000) or .025. If the community has a population of 100, the method to calculate the expected number of cancers in the community would be to multiply the population of the community, 100 by .025, the proportion seen in the reference population (100 x .025). The expected number of cancers in the community would be 2.5.

The observed cases are compared to the expected cases. The relationship between the observed and the expected cases is called a Standardized Incidence Ratio (SIR). If the observed number of cases is the same as the expected number of cases, the SIR is 1.0 and the community has neither a measurable increased nor decreased cancer incidence. If the observed number of cancer cases is lower than the expected number, then the SIR is less than 1.0 and the community has fewer cases than expected. If the observed number of cancer cases is higher than the expected number of cases, the SIR is greater than 1.0 and it is possible that the community is experiencing a greater than expected rate of cancer.

**Standardized Incidence Ratio**

**Standardized Incidence Ratio**

**Standardized Incidence Ratio**

**What It Means**

**What It Means**

**What It Means**

= 1.00

**Standardized Incidence Ratio**

= 1.00

The number of cases in the community is neither higher nor lower than what would be expected based on the number of cases in the reference population.

**What It Means**

The number of cases in the community is neither higher nor lower than what would be expected based on the number of cases in the reference population.

< 1.00

**Standardized Incidence Ratio**

< 1.00

The number of cases is lower than would be expected based on the reference population. It is possible that the community is experiencing less cancer than would be expected.

**What It Means**

The number of cases is lower than would be expected based on the reference population. It is possible that the community is experiencing less cancer than would be expected.

> 1.00

**Standardized Incidence Ratio**

> 1.00

The number of cases is greater than would be expected based on the reference population. It is possible that the community is experiencing more cancer than would be expected.

**What It Means**

The number of cases is greater than would be expected based on the reference population. It is possible that the community is experiencing more cancer than would be expected.

A SIR of 1.5 indicates a 50% increase in cases over what was expected, and .90 indicates a 10% decrease in cases. Interpretation of the SIR depends not only on its size, but also on its stability. A SIR based on a few cases is considered to be unstable, or more prone to chance, than would be the same SIR calculated from a large number of cases.

For example, a SIR of 2, which indicates a 100% increase in observed cancers over expected cancers, would occur if you expected 1 case and observed 2 cases; it would also occur if you expected 100 cases and observed 200 cases. However, in the first instance the SIR is based on 1 excess case, whereas, in the second instance, the SIR is based on 100 excess cases. Both represent a 100% increase, but the latter is far more stable.

Before the conclusion that a community is experiencing an increase or decrease in cancer can be accepted, it is necessary to determine whether the increase is statistically significant. Statistical significance takes into account, by means of a statistical test, variations in the populations and the numbers that are being compared. The SIR assumed that the populations were the same, whereas statistical significance examines the variability in the data.

When an observation is determined to be statistically significant, the association is acknowledged to be due to more than chance alone. Although this does not determine causation, it does indicate that an apparent increase in the rate is probably not due to chance alone. Scientists generally assume statistical significance when there is less than a 5% chance (probability) that a particular outcome occurred by chance alone.

One way of looking at chance is in terms of flipping a coin. Since one has a 50/50 chance of getting either heads or tails in one flip of a coin, one would expect in 10 flips of a coin to get 5 heads and 5 tails. However, it is possible to get 7 heads and 3 tails, 4 heads and 6 tails, or even 10 heads and no tails, purely by chance. The test for statistical significance helps to sort out how often one would expect two things to occur together due to chance alone. Data are said to be statistically significant when the occurrence by chance alone is found to be small.

Since the data used to determine the SIR is a product of estimates, the SIR is an average or an estimate. The Atrue@ ratio is the exact ratio that would be found if we knew exactly how many cases to expect in the population (because a larger reference population is used to estimate the expected occurrence of cancer in a smaller population) and exactly how many cases occurred (since every case of cancer may not have been recorded).

Since it is not possible to determine the true ratio, a range is defined around the SIR and is used to estimate the true ratio. Within this range, it is possible that any number could be the true ratio. Statistical significance is determined by looking at the range around the SIR. This range is referred to as a confidence interval (CI). A 95% confidence interval (95% CI) is a range of numbers around the SIR in which the true ratio falls 95% of the time.

To determine statistical significance, a 95% confidence interval is developed for the SIR. A 95% CI which includes 1, such as (0.96 – 1.13) is not statistically significant because, it is possible that the true ratio could be any number in that range, including 1. A 95% CI which does NOT include 1, such as (1.10 – 1.25), is considered to be statistically significant because it is unlikely that the true ratio would be less than or equal to 1 strictly because of chance. As with the SIR, the stability of the confidence interval must also be evaluated. A narrow confidence interval, (1.5 – 1.7), indicates that the calculated SIR is fairly close to the true SIR for the population and is considered to be stable. A wide confidence interval, (1.1 – 5.7), indicates that the true SIR could be much higher or lower than the calculated SIR and is considered to be unstable. In this situation, even though the SIR is determined significant, the large confidence interval indicates a large amount of variability in the data. Therefore, results should be interpreted with caution.

For example: if 45 cases of breast cancer are observed and 50 cases of breast cancer are expected, the SIR would be 45/50, or 0.90 (a number less than 1 which means there is not an increase in cancer incidence). However, if 50 cases of breast cancer are expected, and 60 cases of breast cancer are observed the SIR would be 60/50, or 1.20 – a number greater than 1, which means the community might be experiencing a greater rate of cancer than is expected. To determine whether the increase is statistically significant, look at the confidence interval around the SIR. If the 95% confidence interval is (.89 – 1.32), the increase is not statistically significant because it is possible that the true ratio could be any number in that range, including .96 to 1. If, on the other hand, the 95% confidence interval is (1.20 – 1.32), the increase is statistically significant because the confidence interval does not include 1, making it unlikely that the true ratio would be less than or equal to 1 strictly because of chance.