SECTION 4
DISCUSSION
One of the primary purposes of the National Exposure Registry (NER) is to determine if there is an excess reporting of adverse health conditions for registrants when compared with national norms. A specific goal of each subregistry is to obtain, maintain, disseminate, and analyze longitudinal data; that is, data collected on the same people over time that have documented exposure to a specific chemical. To date, this goal has been pursued for the TCE Subregistry by the collection of Baseline and Followups 1, 2, and 3 data. The results of the statistical analyses of Followup 1 data are reported and discussed in this document.
The complete TCE Subregistry comprises 4,926 registrants (495 controls from a previous epidemiological study conducted at one of the sites, 4,431 exposed) from 15 sites in 5 states. The Baseline report included the 4,280 (4,041 living, 239 deceased) exposed registrants from 13 sites in 3 states (Illinois, Indiana, and Michigan). The registrants had met the eligibility criterion for Registry participation: during the period of exposure they had resided and used the water at an address with a documented contaminated water supply (private well) for more than 30 consecutive days. The source of contamination varied from site to site, as did the levels and numbers of contaminants other than TCE.
The results contained in the Baseline report (1) were restricted to the 4,154 registrants reporting race as white (3,915 living, 239 deceased). At Followup 1, of the 3,915 registrants alive at Baseline, 3,733 (3,471 living and 262 deceased) were retained on the TCE Subregistry of the NER. The same basic health and demographic information that was collected at Baseline was collected at Followup 1 data collection; information was collected on 3,433 living registrants. The participation rate at Baseline was 98% overall; at Followup 1, 90% of the registrants were retained on the subregistry. About 3% of the loss was due to refusal by registrants to participate. However, most losses were due to ATSDR either being unable to locate some Baseline registrants or being unable to contact the registrant for an interview even though registrants were located.
As discussed in the Baseline report (1), TCE Subregistry and NHIS data were both self-reported. As also discussed in the Baseline report, the response rate of the TCE Subregistry members could possibly have been influenced by the their knowledge of and concern for possible health consequences as a result of the TCE exposure; these factors must be taken into consideration in interpreting the statistical results. The Subregistry and NHIS questions about health conditions shared important similarities, but differed in two ways: (1) restrictions on the source of diagnosis and (2) the wording of some of the questions on health conditions.
Concerning the restrictions on the source of diagnosis, the TCE Subregistry questions, on both the Baseline and Followup questionnaires, specified that the source of diagnosis must be a "physician or other medical provider" (as opposed to self-diagnosed). This limitation was added to each subregistry question about health conditions in an effort to minimize overreporting by registrants whose health awareness might have been heightened by activities occurring at their sites (for example, ongoing litigation), as well as by general public awareness and media coverage of the exposures. The inclusion of this qualifier had the potential for reducing the rates of reporting for the Subregistry when compared with the NHIS rates, all other factors being equal. Of particular interest were the reporting rates for conditions that more commonly would be self-diagnosed--such as asthma, arthritis, hearing impairment, skin rashes, and respiratory allergies--and would predictably be reported at a higher rate by the NHIS participants. Indeed, this was the case; at both Baseline and Followup 1, the reporting was statistically significantly increased for the NHIS population for these conditions, with the exception of skin problems. The comparability of the wording of TCE Subregistry and NHIS health conditions was addressed in Section 3. The implications of dissimilar wording for a health condition must be considered in interpreting statistical differences found in the reporting rates. ATSDR-funded research is currently being carried out to better assess the impact of these instrument differences on reporting rates.
Site-by-site comparisons in terms of health outcomes are not provided in this report. However, it is important to compare the similarities of registrants from site to site in terms of demographic, occupational, and environmental data; this comparison serves as an indicator or check for any aberrant site that might have skewed the data. As was discussed in the Baseline report (1), the sites were comparable in terms of sex ratios and occupational exposure. There were slight differences for the average age and educational attainment; however, these differences were small and would not have been expected to have any marked influence on the rate of reporting for health outcomes. At Followup 1, the nonresponse rates and patterns were similar over sites; therefore, no new differential site biases were introduced in Followup 1.
ENVIRONMENTAL DATA
As was discussed in Section 2, the environmental data available were collected and evaluated, but the data had limitations that merit reiteration. The samples were not taken for the purpose of quantifying exposures over time; rather, they were taken for the purpose of verifying contamination. Regardless, the data served the Subregistry's purpose in that it established that exposure to TCE, in water used for household purposes, had occurred at site addresses (that is, individual residences). Most of the environmental data represent one-time samples of well water at site addresses. These samples might or might not have been taken when TCE and other chemical exposures were at the maximum level; site histories suggest that the samples probably did not represent the peak level.
Analyses of the environmental data presented in the Baseline report (1) focused on whether there was an association between the levels of exposure--both for TCE and other chemicalsand the rates of reported health outcomes. The results of these analyses, after controlling for the effects of sex, age, cigarette smoking history, and occupation, showed that there was not a consistent increase of health symptom reporting with increased levels of exposure and length of exposure; however, some statistically significant associations were found--cumulative chemical exposure with respiratory allergies, length of exposure with hearing impairment, and maximum TCE quartiles with stroke). Given the limitations of the environmental data, associations between health conditions and exposure levels were difficult to establish. Future ATSDR and other studies are exploring additional methods for obtaining environmental information needed for the investigation of dose-response relationships.
HEALTH OUTCOMES
The comparisons of TCE Subregistry Followup 1 and NHIS data on reported health conditions revealed several statistically significant differences. These findings are summarized in Table 4-1, along with the results of the statistical findings from the Baseline data. It should be noted that the age intervals have been advanced 1 year from the Baseline report, thus maintaining the same groupings as at Baseline. Also of note is that the large sample size and setting the Type I error (the level) at 0.01 served to help control the likelihood of false negative or false positive results. The results of the Baseline report, while discussed in this report and compared with the Followup 1 results, were not known to the registrants at the time of Followup 1 data collection and, therefore, could not have influenced registrant reporting.
Speech Impairment
Reporting of speech impairment ("currently have" is the time frame for reporting) was statistically significantly elevated for the 0 to 9 years of age group, both sexes, at Baseline. This elevation was observed (but not statistically significant) for females, but not males, at the first followup. The reporting rate for this age group decreased from 4.3% at Baseline to 2.5% at Followup 1.
Because of the statistically significant increase in reporting and the relevant published toxicological information (1) supporting the biological plausibility of an association of TCE exposure and this outcome, a follow-up study is being funded by ATSDR to explore the possible association between excess speech and hearing impairments and TCE exposure for the group less than 10 years of age at Baseline.
Hearing Impairment
At Baseline, hearing impairment was reported at a statistically significantly lower rate by registrants 25 years of age and older (a result predicted because of the necessary physician confirmation for TCE Subregistry respondents but not for NHIS respondents), but at a statistically significantly higher rate for those 9 years of age or younger. Similar to Baseline results, at Followup 1 statistically significantly decreased reporting results were observed for the registrants 25 years of age and older, and also for those 11 through 24 years of age. Although the statistically significant elevation of subregistry rates for those under 10 years of age which was observed at Baseline was not present in the Followup 1 data, the risk ratio for this age group at Followup 1 was markedly higher than for the other age groups.
For several reasons--the close association between speech and hearing impairments, the increased reporting rate at Baseline, and, as with speech impairment, the literature support of biological plausibility (1)--hearing impairment is included as part of the ATSDR follow-up study of speech and hearing impairments.
Table 4-1.--Summary results of TCE Subregistry and National Health Interview Survey comparisons for Baseline and Followup 1.
Disease Category |
Age Groups (Years) | ||||||||||||||||||
| 1-10 | 11-18 | 19-25 | 26-35 | 36-45 | 46-55 | 56-65 | 66+ | All | All | ||||||||||
| M | F | M | F | M | F | M | F | M | F | M | F | M | F | M | F | M | F | ||
Speech impairment |
|||||||||||||||||||
| Baseline | X | X | -- | -- | |||||||||||||||
| Followup 1 | |||||||||||||||||||
Hearing impairment |
|||||||||||||||||||
| Baseline | X | X | R | R | R | R | R | R | R | R | R | R | |||||||
| Followup 1 | R | R | R | R | R | R | R | R | R | R | R | R | R | R | |||||
Hypertension |
|||||||||||||||||||
| Baseline | |||||||||||||||||||
| Followup 1 | -- | R | R | ||||||||||||||||
Stroke |
|||||||||||||||||||
| Baseline | -- | -- | -- | -- | -- | -- | X | X | X | X | X | X | |||||||
| Followup 1 | -- | -- | -- | -- | |||||||||||||||
Liver problems |
|||||||||||||||||||
| Baseline | -- | -- | -- | -- | -- | -- | -- | -- | -- | -- | -- | X | |||||||
| Followup 1 | -- | X | X | ||||||||||||||||
Anemia and other blood disorders |
|||||||||||||||||||
| Baseline | X | -- | X | X | X | X | X | X | |||||||||||
| Followup 1 | -- | X | X | X | |||||||||||||||
Diabetes |
|||||||||||||||||||
| Baseline | -- | -- | -- | -- | X | X | |||||||||||||
| Followup 1 | -- | X | X | X | |||||||||||||||
Kidney disease |
|||||||||||||||||||
| Baseline | -- | -- | X | -- | |||||||||||||||
| Followup 1 | |||||||||||||||||||
Table 4-1.--Continued.
Disease Category |
Age Groups (Years) | ||||||||||||||||||
| 1-10 | 11-18 | 19-25 | 26-35 | 36-45 | 46-55 | 56-65 | 66+ | All | All | ||||||||||
| M | F | M | F | M | F | M | F | M | F | M | F | M | F | M | F | M | F | ||
Urinary tract disorders |
|||||||||||||||||||
| Baseline | -- | X | -- | -- | X | X | X | X | X | X | X | ||||||||
| Followup 1 | -- | X | X | X | X | X | X | X | X | X | X | X | X | ||||||
Skin rashes |
|||||||||||||||||||
| Baseline | X | ||||||||||||||||||
| Followup 1 | X | X | X | ||||||||||||||||
Asthma, emphysema |
|||||||||||||||||||
| Baseline | R | ||||||||||||||||||
| Followup 1 | |||||||||||||||||||
Mental retardation |
|||||||||||||||||||
| Baseline | |||||||||||||||||||
| Followup 1 | R | ||||||||||||||||||
Arthritis |
|||||||||||||||||||
| Baseline | R | R | |||||||||||||||||
| Followup 1 | X | R | R | R | R | R | R | R | R | R | |||||||||
Respiratory allergies |
|||||||||||||||||||
| Baseline | R | R | |||||||||||||||||
| Followup 1 | X | R | R | R | |||||||||||||||
Cancers |
|||||||||||||||||||
| Baseline | |||||||||||||||||||
| Followup 1 | X | ||||||||||||||||||
X = Statistically significant differences, TCE Subregistry rate higher.
R = Statistically significant differences, NHIS rate higher.
-- = Insufficient data.
Hypertension
At Baseline, there were no statistically significant differences in rates of reporting for hypertension. In the Followup 1 data, there was statistically significantly higher reporting of hypertension among the NHIS respondents for males 36 through 55 years of age; an unpredicted and unexplainable result. It should be noted that the upper bound of the 99% confidence intervals were very close to 1. Of greater interest, however, is the pattern found in the data: for the age groups in which the TCE Subregistry male participants underreported, the female reporting rates exceeded (although, not statistically so) the NHIS rates. Also of interest is that 9 registrants under the age of 18 years reported positively compared with the 1.9 expected based on NHIS reporting rates.
Reports in literature of hypertension following exposure to TCE are mixed. What is recognized is that hypertension is a common condition associated with diabetes (14,15); this interrelationship needs to be further explored in the context of the total statistical results of the Baseline and Followup 1 analyses.
Effects of Stroke
At Baseline, statistically significantly elevated rates were reported for stroke (in an "ever" time frame) by registrants 35 through 54 years of age and those 64 years of age and older. These statistical significance were not noted at Followup 1. Adding the additional year of information (as discussed in Section 3, the cumulative rate was used for this outcome) actually increased the 35 through 45 years of age group observed to expected ratio from 3.11 at Baseline to 3.80 at Followup 1; however, the elevated rate was not statistically significant. For the 46 through 55 years of age group, the TCE Subregistry rate did not change, but the NHIS reporting rate increased considerably (from 1.75 expected at Baseline to 4.45 at Followup 1). Therefore, it appears that the discrepancies in the statistical significances between Baseline and Followup 1 were the result of statistical aberrations (including zeros in the denominator) rather than changes in TCE Subregistry reporting rates. Overall, the risk ratios by age groups were close to or exceeded Baseline ratios.
As noted in the Baseline report (1), the literature does not document any known association of stroke and TCE exposure. What was pointed out in the Baseline report, however, was that the TCE Subregistry population also had a statistically increased reporting rate for diabetes, which--along with other factors such as smoking--is known to be correlated with the occurrence of stroke.
Liver Problems
The total number of liver problems reported rose from 13 at Baseline to 23 at Followup 1. Nine of the 23 reported cases were males in the age groups 26 through 45 years; no cases were reported for these age groups at Baseline. This was a statistically significant increase in reporting in males 26 through 35 years of age. At Baseline, females 55 to 64 years of age showed a statistically significantly elevated rate of liver problems; the 45 through 54 years of age group was also elevated. At Followup 1, the rate for females 46 through 55 years of age showed a statistically significant elevation; the rate for the 56 through 65 years age group was elevated (actually 1 more case reported than at Baseline), but was not statistically so when compared with the 1990 NHIS rate. It should be noted that the reporting of a liver problem is a rare event and the change of one reported case can alter statistical significance.
The relationship of TCE exposure and hepatotoxicity and the mechanism for this toxicity is widely discussed in the literature (16). The potentiation of the TCE toxicity by other chemicals, such as ethanol, is also widely discussed (17). In further evaluation of the statistical results for this outcome, other potentially influencing factors should be considered.
Anemia or Other Blood Disorders
At Followup 1, females 19 through 25 years of age and 56 through 65 years of age and males 66 years of age and older showed a statistically significantly elevated reporting rate for anemia and other blood disorders. The younger female group and male groups showed a similar elevation at Baseline; the Followup 1 female 56 through 65 years of age group were not statistically significantly elevated at Baseline. Likewise, there were additional groups statistically significantly elevated at Baseline that were not at Followup 1--males 1 through 10 years of age, males and females 36 through 45 years of age, females 46 through 55 years of age, and males 56 through 65 years of age. What were consistent, however, were the increased TCE Subregistry reporting rates across all age groups although the statistical significance varied with varying NHIS rates and the number of positive reports. For example, the reporting rate for males 10 years of age or younger was no longer statistically significant; however, the decrease in the observed-to-expected rate was small--from 3.54 to 3.15.
As noted in the Baseline report (1), in view of the published literature, the association of chronic TCE exposures and anemia is biologically plausible. It must be cautioned, however, that the results reported here do not--similar to the Baseline and other NER results--establish a cause and effect relationship.
Diabetes
The two age-, sex-groups who exhibited an elevated rate of diabetes at Baseline, females 18 through 24 years of age and 45 through 54 years of age, still exhibited this statistically significantly elevated rate at Followup 1. However, another female age group--45 through 65 years--also showed statistically significantly elevated rates at Followup 1.
As noted in the Baseline report (1), few studies are available that explore a possible relationship between chemical exposure and the development of diabetes. The consistent excess reporting by the exposed population needs further exploration--taking into account such concomitant factors as family histories and occupational exposure.
Kidney Disease
None of the age-, sex-groups showed statistically significantly different rates for kidney disease at Followup 1. This was consistent with the results at Baseline with one exception--females 55 through 64 years of age reported statistically significantly higher at Baseline; while elevated, the observed-to-expected ratio (2.22) was not statistically significantly different from 1. What is of note is that the number of positive reports in the 10 years of age or younger group increased from 1 to 5 (the observed-to-expected ratio was 3.156; 99% CI = 0.68, 8.93).
As noted in the Baseline report (1), the literature concerning the association of TCE exposure and kidney disease is mixed; it is an outcome that will be followed closely in future analyses.
Urinary Tract Disorders
Consistent with Baseline (except for the age group 46 through 55 years of age which was not statistically significantly elevated at Baseline but is at Followup 1) statistically significantly elevated rates of reporting for urinary disorders were observed for all female age groups, including those 10 years of age or younger. Statistically significant increases in reporting were found in males at Baseline and Followup 1 for groups 19 through 35 years of age and additionally (although marginally significant) for Followup 1 for groups 11 through 18 years of age and 56 through 65 years of age.
The interpretation of these results is problematic. As discussed in the Baseline report (1), the wording of the NER question allows for broader reporting than the comparable NHIS question; however, the differential reporting by females, which is what would have been expected, mandates continued close scrutiny of the future related findings. The information pertaining to the relationship between TCE exposure and urinary problems is very sparse. The question will be revised for future data collections to make the NHIS and NER wording more parallel.
Skin Rashes, Eczema, or Other Skin Allergies
Three age, sex groups--females 10 years of age or younger, males 11 through 18 years of age, and females 19 through 25 years of age--showed a statistically significantly increased rate of reporting skin rashes at Followup 1. At Baseline, the summary model of the overall rate was assessed for significance, and the overall rate was statistically significantly elevated at Baseline; however, the age, sex groups statistically significantly elevated at Followup 1 were the major contributors to the overall significance at Baseline.
In general, based on the literature, the finding of excess skin rashes, eczema, or other skin allergies in a population exposed to TCE and other volatile organic compounds was not unexpected (1) and will be closely monitored in future data collections.
Asthma, Emphysema, or Chronic Bronchitis and
Other Respiratory Problems Such as Hay Fever
The analysis of Baseline data revealed a statistically significant increase by the NHIS respondents over the registrants in the reporting rate for asthma, emphysema, or chronic bronchitis and for other respiratory problems such as hay fever--an outcome predicted because of the additional physician confirmation needed for NER responses. This difference was no longer significant for Followup 1 data; that is, the reporting rate of the NHIS data, without the physician confirmation qualifier, and the NER reporting rate, with this qualifier, were very similar.
For the outcome other respiratory allergies or problems such as hay fever, there was also a deviation from Baseline: the rate for males in the 10 years of age or younger was statistically significantly increased over the NHIS rate while other sex, age groups (males 10 years of age or younger, females 36 through 45 years of age, and males 36 through 45 years of age) remained significantly statistically greater for the comparable NHIS groups. The statistically significant increases in NER reporting for some sex, age groups reflects both an increase in NER reporting and a decrease in NHIS reporting.
The literature is mixed on the association of TCE and these health outcomes. Previous epidemiological studies (18, 19, 20) have reported associations in exposed communities.
Mental Retardation
Neither age nor sex was a significant predictor for mental retardation for Followup 1 data. Thus, the comparison with NHIS data was made using the overall rate. The overall rate was statistically significantly different, with the NHIS group reporting at a higher rate. Even while it was possible to construct a statistical model and obtain convergence with Followup 1 data (unlike at Baseline), the number of cases of mental retardation reported (in the current time frame) are very few and the interpretation of the statistical significance result found was not straightforward. With so few cases observed (14 at Baseline, 7 at Followup 1), the change of one answer could have had a great effect on the results obtained and statistical significances found.
Arthritis, Rheumatism, or Other Joint Disorders
The results seen for arthritis at Followup 1 and Baseline were consistent overall; as predicted, there was a statistically significant overall lower reporting rate for the TCE Subregistry participants than for the NHIS. For the Followup 1 data, unlike for the Baseline data, the fit of the model was questionable and therefore, the model was not used for further calculations and comparisons; the comparisons were made for the specific age, sex groups. As at Baseline, the rate of reporting for NHIS participants was statistically significantly higher than, or not different from, that of the TCE registrants with one exception--for males 11 through 18 years of age the TCE registrant reporting rate was statistically significantly higher than for the NHIS respondents (7 reported versus 1.7 expected).
The literature contains one reference relevant to the possible association of TCE exposure and this outcome (20). This finding will be further explored with additional follow-up information.
Cancers
The results for Followup 1 registrant reporting rates for health care provider diagnoses or treatment for cancers in the last 12-month time frame indicate increased reporting, when compared to expected values based on NHIS reporting, for female registrants in the 19 through 25 years of age group. At Baseline, a general increase in reporting (using the constructed 12-month time frame rates) was found for the TCE Subregistry members but the increase was not statistically significant. For Follow 1, there two differences from Baseline to be noted--first, the timeframes were the same for the NHIS and NER questionnaires and derived rates were not necessary for the TCE Subregistry file and second, the NHIS comparison file reporting rates were decreased from the NHIS file used for the Baseline comparison. Both of these differences have the potential to impact the statistical results. The total number of cancers increased from 55 at Baseline to 64 at Followup 1 for the TCE registrants; the expected number, based on the NHIS reporting rates, deceased from 45.6 to 35.2.
Results of studies of the potential carcinogenic properties of TCE, as discussed in the Baseline report (1), have varied. Based on the literature and on the results of Followup 1 and the Baseline analyses, it appears prudent to further evaluate the number and types of cancers being reported and validate the findings with medical records.
SUMMARY
In this section, statistical results comparing reporting rates for multiple health outcomes by a TCE-exposed population with national norms (NHIS) were detailed and discussed. As detailed in the Baseline report (1), a survey of the literature suggests that there might be associations between TCE exposure and some of the health outcomes that were reported in excess, both at Baseline and Followup 1, by the subregistry participants. As also noted in the Baseline report, the reported literature has many limitations--case reports of human poisonings and occupational studies usually involve exposure levels much higher than those reported in environmental exposures; high-dose animal studies might not be relevant to humans; and human health studies often lack sufficient exposure characterization, controls for important confounding factors, and sample sizes large enough to investigate low-dose effects. These and other limitations must be considered. This report does not support a cause-and-effect relationship between TCE exposure and human health outcomes, based either on the Baseline data or Followup 1 data. However, interesting areas for further investigation do present themselves.
Agency for Toxic Substances and Disease Registry, 1825 Century Blvd, Atlanta, GA 30345
Contact CDC: 800-232-4636 / TTY: 888-232-6348
