National Academies Press: OpenBook

Enhancing Data Systems to Improve the Quality of Cancer Care (2000)

Chapter: The Data Infrastructure for Health Services Research

« Previous: How are Healthcare Systems Monitoring Quality Today?
Suggested Citation:"The Data Infrastructure for Health Services Research." Institute of Medicine and National Research Council. 2000. Enhancing Data Systems to Improve the Quality of Cancer Care. Washington, DC: The National Academies Press. doi: 10.17226/9970.
×

4

The Data Infrastructure for Health Services Research

This chapter describes data resources for cancer-related health services research, the study of the structure, processes, and effects of healthcare services. Relative to funding for basic cancer research, support for research in this field is quite modest (IOM, 1999a). Even so, the agencies supporting health services research and investigators in this field have developed innovative methods to enhance available data resources through linkages, special studies, the establishment of research consortiums, and new data collection initiatives. This chapter begins with a description of selected programs that take advantage of current data systems to further cancer-related health services research, and concludes with a discussion of the limitations of federally funded surveys when applied to quality-of-care issues.

LINKAGE OF CANCER REGISTRIES TO ADMINISTRATIVE DATA

A great deal has been learned about the quality of cancer care from studies that link two or more complementary data sources. The linkage of cancer registry data to insurance claims databases, for example, has provided evidence of significant geographic variations in care and has suggested that care within certain HMOs for certain cancers is as good as, or superior to, the care provided in fee-for-service plans (Potosky et al., 1997; Riley et al., 1999).

Registry data contain useful measures of severity of cancer (e.g., cancer stage) and date of diagnosis but may lack complete information on treatment and outcomes. Claims-based data may lack certain diagnostic information but in

Suggested Citation:"The Data Infrastructure for Health Services Research." Institute of Medicine and National Research Council. 2000. Enhancing Data Systems to Improve the Quality of Cancer Care. Washington, DC: The National Academies Press. doi: 10.17226/9970.
×

clude detailed information on the cost and use of medical services (Fleming and Kohrs, 1998). Claims data are often accessible, routinely collected, and represent the utilization experience of a large number of patients. Their limitations, however, include coding misspecification and errors, incomplete listing of extant disease (e.g., listing only a limited number of diagnoses on hospital discharge files), and difficulties in distinguishing incident from prevalent conditions (e.g., whether a condition listed arose during a hospitalization or was preexisting).

One of the most fruitful linkages for cancer care assessment is that of the Surveillance, Epidemiology, and End Results (SEER) cancer registries to claims records in Medicare's administrative database (Moulton, 1998). This is a collaborative effort of the National Cancer Institute (NCI), the SEER registries, and the Health Care Financing Administration (HCFA) to create a large populationbased source of information for cancer-related epidemiologic and health services research (Potosky et al., 1993). The SEER registries are located in 11 geographic areas and 3 supplemental registries that include 14% of the U.S. population (SEER registries are described more fully in Chapter 5). The Medicare utilization data (claims) cover stays in institutions (i.e., hospitals and skilled nursing facilities), physician and lab services, hospital outpatient visits, and home health and hospice use. Information on noncovered services such as prescription drugs, and long-term care is not included. The currently available linked file includes all Medicare data through 1998 for persons diagnosed with cancer in 1996.

Matching a case from the cancer registry to claims in the Medicare files is performed using a computer program that applies an algorithm to determine whether records from the two sources represent the same individual based on available identifying information (i.e., social security number, name, birth date, gender). Of persons age 65 and older reported by the SEER registries, 93% were matched to the Medicare master enrollment files. A failure to match may occur if the patient identified in the registry is not a Medicare beneficiary (e.g., an estimated 3% of the elderly do not qualify for Medicare) or errors are made in recording identifying information.

Once a match is established, Medicare claims are extracted. The database includes claims for beneficiaries receiving fee-for-service care and excludes information about care provided to individuals cared for in HMOs, those in the Department of Veterans Affairs (VA) medical system, and those whose care is paid for exclusively with private health insurance. Before release, all information that can identify an individual is stripped from the files. Data files are made available for research on a limited basis through an application process. Representatives from NCI, the SEER registries, and HCFA review each proposal to ensure that the research does not compromise the confidentiality of patients or medical care providers in SEER areas. Researchers who use the SEER-Medicare files must sign agreements to abide by strict confidentiality rules.

The SEER–Medicare data offer an opportunity to examine patterns of care prior to the diagnosis of cancer, during the period of initial diagnosis, and during

Suggested Citation:"The Data Infrastructure for Health Services Research." Institute of Medicine and National Research Council. 2000. Enhancing Data Systems to Improve the Quality of Cancer Care. Washington, DC: The National Academies Press. doi: 10.17226/9970.
×

long-term follow-up. Topics that can be addressed with the linked database include patterns of care for specific cancers, the use of health services, and the costs of treatment (Deleyiannis et al., 1997; Du et al., 1999; Lu-Yao et al., 1996; Potosky et al., 1999; Riley et al., 1999). Longitudinal surveillance of the health care of persons with cancer is another potential use of this linked file. These data can be used to assess health care directed toward the prevention of disease or disability, as well as the restoration or maintenance of health (Edwards, 1997). A control sample of individuals who do not have cancer is available so that comparisons can be made, for example, on healthcare costs for individuals with and without cancer (NCI/SIG, 1999; Warren et al., 1999). Active projects using the linked SEER–Medicare database include analyses of:

  • total lifetime payments for elderly cancer patients,

  • differences in patterns of care and cancer survival between HMOs and fee-for-service providers,

  • breast cancer treatment patterns and trends,

  • prostate cancer detection practices,

  • trends and variations in initial treatment for early-stage prostate cancer, and

  • hospice use among beneficiaries with colorectal and lung cancer.

State cancer registry data have also been linked to Medicare claims (Hillner, 1995; Smith et al., 1995), private insurance claims (Hillner, 1997), and hospital discharge data (Ayanian et al., 1993; Polednak et al., 1996) to assess quality of care (see a listing of selected state registry-based quality studies in Appendix C). The inclusion of the social security number on reports to the National Program of Cancer Registries (NPCR) state registries facilitates successful linkages.

One study under way will test the use of multiple data linkages to assess the quality of cancer care (Box 4.1). If successful, it could foster public reporting of risk-adjusted quality measures for health plans or providers, provide techniques to develop benchmarking standards for internal quality improvement, and establish a standard for surveys of patients' appraisals of care.

CANCER REGISTRIES AS A SAMPLING FRAME FOR SPECIAL STUDIES

Another mechanism used to assess quality of care is to use cancer registries as a sampling frame for special studies. The registries often have near complete ascertainment of incident cases of cancer but lack additional information needed for quality studies (e.g., information on outpatient treatments). Data elements needed to answer specific questions can be obtained through medical chart abstraction and/or patient survey from a representative sample of cases.

NCI has, since 1988, conducted a series of patterns-ofcare studies using this model. NCI's intent is to describe the dissemination of state-of-the-art can-

Suggested Citation:"The Data Infrastructure for Health Services Research." Institute of Medicine and National Research Council. 2000. Enhancing Data Systems to Improve the Quality of Cancer Care. Washington, DC: The National Academies Press. doi: 10.17226/9970.
×

cer treatment and explanatory factors for variations in patterns of care. Currently, data are being collected on cases diagnosed in 1998 within SEER registries with cancer of the prostate, corpus uteri, pancreas, and chronic lymphocytic

BOX 4.1 Using State Cancer Registries Linked to Other Data to Assess the Quality of Cancer Care

An ambitious study under way in California is attempting to capitalize on the strengths of State cancer registry data while compensating for their weaknesses. Investigators at the Harvard Medical School are collaborating with the California Cancer Registry to assess the following process and outcome measures for individuals with colorectal cancer:

  • stage at diagnosis,

  • timeliness of treatment,

  • provision of recommended chemotherapy and radiation therapy,

  • patient reported quality of care, and

  • survival.

As a first step, the investigators will contact all the physicians who cared for patients with colorectal cancer that were reported to the registry in 1996–1997. Information reported to the cancer registry will be verified and some additional information about the patient's care will be collected, for example, the types of drugs used and the timing of treatment. If the registry data is found to be inaccurate, they will assess whether certain types of hospitals or areas are prone to poor reporting.

Next, the cancer registry data will be linked to a number of data files:

Type of File

Information Provided

Medicare enrollment data (beneficiaries age 65 and older)

Type of health plan and the name of the plan—most people with colorectal cancer are elderly

Hospital discharge abstracts

Comorbid conditions identified at the time of surgery

Physician specialty data

Training of the physician who reported the cancer

U.S. Census data

The patient's area of residence, for example, the level of poverty in the neighborhood

Suggested Citation:"The Data Infrastructure for Health Services Research." Institute of Medicine and National Research Council. 2000. Enhancing Data Systems to Improve the Quality of Cancer Care. Washington, DC: The National Academies Press. doi: 10.17226/9970.
×

When all of the files are linked, the investigators will have a fairly comprehensive database with which to evaluate the correlates of good care, for example, whether care was received in a managed care vs. a fee-for-service system, and the type of physician providing care. To gain the patients' perspective on care, the investigators plan to interview a sample of patients 6 months after their diagnosis was made regarding their experience with care (e.g., coordination and continuity of care, physical comfort, trust in doctors and health plans). Aspects of the patients' quality of life will also be assessed, for example, their functional and emotional well-being.

SOURCE: Ayanian, 1999.

leukemia.1 Participating cancer registries reabstract hospital records for additional data and verify data regarding therapies with physicians. An NCI patterns-of-care study typically takes 5 years from conception to publication of results.

Sampling from SEER registries is also central to NCI's special studies, a mechanism to collect in-depth data beyond that routinely collected for cancer registration (NCI/SIG, 1999). With a time frame of 1 to 2 years, these studies can provide a rapid response to questions of national importance. A SEER study, for example, was used to obtain improved estimates of the risks associated with tamoxifen during a controversial period of the Primary Prevention Trial of Breast Cancer. One of the larger special studies, the Prostate Cancer Outcomes Study, is addressing health-related quality-of-life issues among the increasing population of men identified with prostate cancer who have had radical prostactectomy. As part of this study, a cohort of 3,500 men with prostate cancer was identified from six SEER registries, and health-related quality of life is being measured at 6, 12, and 24 months and at 5 years following diagnosis. Community practice patterns are also being assessed, for example, variations in diagnostic and treatment interventions. The experience with the prostate cancer special study indicates that incident case cohorts can be successfully identified and tracked longitudinally using an existing data collection infrastructure (i.e., the SEER cancer registries) (Potosky, 1999; Potosky et al., 1999).

The American Cancer Society (ACS) is sampling cases from a few state cancer registries as part of a pilot test for two large population-based surveys of cancer survivors (Baker, 1999). The first is a planned 10-year prospective study of survivors enrolled within the first year after diagnosis of any one of the 10

1  

Samples of cases from SEER were obtained in 1988, 1989, 1990, 1991, 1995, and 1996 to assess the following cancers: in situ and early-stage breast cancer, colorectal, ovarian, urinary bladder, melanoma, non-small-cell lung, head and neck, cervix, childhood brain stem, and other childhood cancers.

Suggested Citation:"The Data Infrastructure for Health Services Research." Institute of Medicine and National Research Council. 2000. Enhancing Data Systems to Improve the Quality of Cancer Care. Washington, DC: The National Academies Press. doi: 10.17226/9970.
×

most common cancers.2 The major aim of the survey is to examine the behavioral, psychosocial, treatment, and support factors that influence quality of life and survival of cancer patients. Plans are to extend the study to other states that have adequate cancer registration and an interest in participating, with the ultimate goal of enrolling a sufficient number of subjects to provide state-level estimates (i.e., up to 100,000 subjects nationwide). Difficulties encountered during the pilot phase of the study have included:

  • the lack of rapid case ascertainment mechanisms to identify cases early enough to administer a survey within 1 year of diagnosis,

  • shortages of resources and staff within the registries,

  • subjects unavailable for study because of involvement in other research studies, and

  • adherence to physician and patient consent legal requirements being labor intensive (Baker, 1999).

The second survey is a cross-sectional study of 6,000 long-term survivors (i.e., those who are 2, 5, and 10 years beyond diagnosis) of 6 cancers (prostate, breast, colorectal, bladder, melanoma, uterine). The study design calls for 1,000 respondents for each type of cancer. The original study design included 15-year survivors, but relatively few registries were established in 1983 or earlier and had complete data necessary to identify 15-year survivors.

DEVELOPING HEALTH SERVICES RESEARCH CONSORTIUMS

Many studies of cancer care quality exclude members of managed care organizations because such plans often do not have encounter data available (e.g., individual claims for visits or services). Such plans, however, cover the majority of privately insured Americans and have internal data systems available on the care of their members. A new initiative of the NCI, the Cancer Research Network (CRN), will encourage the expansion of collaborative cancer research among healthcare provider organizations that are oriented to community care; have access to large, stable, and diverse patient populations; and are able to take advantage of existing integrated databases that can provide patient-level information relevant to research studies on cancer control and to cancer-related population studies. Beginning in 1999, NCI funded the first CRN—a consortium of 10 large, not-for-profit, research-oriented HMOs. The CRN will conduct four main projects (Brown, 1999):

2  

The cancer sites include: prostate, female breast, lung, colorectal, urinary bladder, non-Hodgkin's lymphoma, skin melanoma, uterine, kidney, and ovarian.

Suggested Citation:"The Data Infrastructure for Health Services Research." Institute of Medicine and National Research Council. 2000. Enhancing Data Systems to Improve the Quality of Cancer Care. Washington, DC: The National Academies Press. doi: 10.17226/9970.
×
  1. a study of tobacco control policies and programs and their relationship to patient smoking cessation rates, and an analysis of healthcare smoking-related costs;

  2. a study of late-stage breast and invasive cervical cancer cases to elucidate the patient, provider, and system factors that contribute to preventing advanced disease;

  3. a study of the effectiveness of the commonly used strategies of frequent mammography or prophylactic mastectomy, to prevent fatal breast cancer among women at increased risk for breast cancer; and

  4. a test of methods for increasing the participation of HMO patients in cancer clinical trials.

The CRN infrastructure will include a data-coordinating center and expert teams to provide scientific input in the areas of biostatistics, health economics, survey measures, pharmaco-epidemiology, genetics, clinical trials management, and survivorship issues.

FEDERAL HEALTH SURVEYS AND DATA

The federal government spends a considerable amount on statistical activities related to health—nearly a billion dollars alone on direct funding of major statistical programs within the Department of Health and Human Services (OMB, 2000). The results from surveys and other data collection activities provide national estimates of such health indicators as the prevalence of health conditions, the use of healthcare services, and healthcare expenditures. Federal agencies also support methodological research that has fostered the development of standardized survey instruments (e.g., patient satisfaction with health care) and techniques to improve data processing and analysis. Examples of federal health surveys include the National Health Interview Survey, the National Ambulatory Medical Care Survey (NAMCS), the National Hospital Discharge Survey, and the Medical Expenditure Panel Survey. Some surveys are conducted for certain populations (e.g., the Medicare Current Beneficiary Survey) while others are targeted to specific health conditions (e.g., the AIDS Cost and Services Utilization Study). National surveys have been invaluable in estimating the prevalence of cancer risk behaviors (e.g., smoking) and use of preventive health services (e.g., mammography use) but have not been as useful in treatment-related quality-of-care studies.

Federal surveys conducted of individuals are often very large, including members of as many as 50,000 households. Even so, the incidence of cancer is estimated to be under 1%, making it difficult to identify large numbers of recently diagnosed cases of any particular type of cancer. Household surveys exclude residents of institutions and therefore miss individuals with cancer who are in nursing homes, hospices, or other facilities. There are also limitations in self-reports of cancer. Evidence suggests, for example, that many individuals do not accurately report the occurrence of cancer or the type of cancer diagnosed (Bergmann et al., 1998; Chambers et al., 1976). When national surveys of

Suggested Citation:"The Data Infrastructure for Health Services Research." Institute of Medicine and National Research Council. 2000. Enhancing Data Systems to Improve the Quality of Cancer Care. Washington, DC: The National Academies Press. doi: 10.17226/9970.
×

healthcare providers or facilities (e.g., NAMCS) are conducted, similar problems occur, such as accruing a sufficient sample of individuals with cancer, obtaining sufficient information on stage of illness and comorbidity, and determining the indication for a procedure (e.g., curative vs. palliative surgery).

Investigators at ACS have analyzed the National Hospital Discharge Survey from 1988 to 1995 to describe patterns of use of inpatient surgical procedures for treating cancers of the lung, colon/rectum, prostate, and female breast, by age, gender, race, and geographic region (Wingo, 1999). The results are useful in assessing general trends in service use and in generating hypotheses on apparent disparities in use, but they are difficult to interpret because of the lack of information on cancer stage and indication for the procedures. In general, national health surveys are extremely useful in gauging progress toward goals established in the area of cancer control and prevention but have limited applications in assessing other aspects of the quality of cancer care.

Some surveys include a sufficient sample of cancer patients to make robust analyses possible. Health services researchers interested in end-of-life care issues, for example, have used two surveys sponsored by the National Center for Health Statistics (NCHS), the National Home and Hospice Care Survey and the National Mortality Followback Survey (NMFS) (www.cdc.gov/nchs). As part of the NMFS, nearly 23,000 1993 death certificates were sampled and next of kin interviewed on where the decedent's death occurred, use of health care during the last year of life, unmet health needs, and the quality of the last year of life (e.g., functional limitations, use of pain medication).

A valuable data resource with which to assess the quality of hospital care is the Healthcare Cost and Utilization Project (HCUP) of the Agency for Healthcare Research and Quality (AHRQ). HCUP includes two databases for health services research:

  • the Nationwide Inpatient Sample includes inpatient data from a national sample of about 1,000 hospitals, and

  • the State Inpatient Database covers inpatient care in community hospitals in 22 states that represent more than half of all U.S. hospital discharges.

The uniform data in HCUP make possible comparative studies of the use and cost of hospital care, including the effects of market forces on hospitals and the care they provide, variations in medical practice, and the use of services by special populations (www.ahrq.gov/data/hcup). Examples of cancer-related health services research using HCUP include an analysis of hospital characteristics associated with breast-conserving surgery (Johantgen et al., 1995) and a study of geographic variations in sphincter-sparing surgery rates in the treatment of rectal cancer (Morris, 1999). Analyses of the association between volume of services and short-term outcomes for specific procedures are also possible because the HCUP databases include all discharges for a given hospital, not just a sample of discharges.

Suggested Citation:"The Data Infrastructure for Health Services Research." Institute of Medicine and National Research Council. 2000. Enhancing Data Systems to Improve the Quality of Cancer Care. Washington, DC: The National Academies Press. doi: 10.17226/9970.
×

SUMMARY

One of the most productive strategies health services researchers have used to assess the quality of cancer care has been to link cancer registry data to either administrative claims records or hospital discharge files. The data sources are often complementary—cancer registry data contain important information on diagnosis and cancer stage but may not record complete information on treatment that occurs outside the hospital. Administrative data may lack the diagnostic information but record a patient's treatment encounters. Linked data sets are not without problems. Administrative records may have treatments miscoded, comorbidity data needed to adjust results may be limited, and data elements necessary for complete linkage may be absent. Nevertheless, such linkages have allowed researchers to study variation in cancer care and to make comparisons across systems of care. A large study being conducted in California will provide information on the quality of registry treatment data, as well as maximize the potential of linkage using the cancer registry, hospital discharge data, and claims data from the Medicare program.

Many cancer registries are achieving nearly complete levels of case ascertainment, making them valuable as sampling frames for targeted special studies. Here, cancer registry staff may be asked to gather from medical charts clinical information to supplement that obtained for routine registry purposes. With appropriate resources, special studies can be launched relatively quickly in response to a specific research question. The SEER program has conducted a number of special studies, including a recent study of quality-of-life issues among men with prostate cancer following prostatectomy. The American Cancer Society is piloting two cancer survivorship surveys using state registries as sampling frames. Some state laws regarding confidentiality and consent have made timely access to research subjects difficult (e.g., requiring consent from the patient and the attending physician). Furthermore, Institutional Review Boards within hospitals can take several months to approve research projects.

The majority of privately insured Americans receive care within managed care organizations, but data on their care are often difficult to obtain because individual claims are usually not filed for each encounter. A number of plans have internal information systems and a population-based orientation to health care, making them ideal partners for research. The NCI has developed a consortium of such large managed care plans to collaborate on research.

Many of the federally sponsored national surveys can provide important descriptive information relevant to cancer, for example, use of health services and trends in service use. However, such surveys generally have limitations for cancer-related health services research because of the relatively rare occurrence of incident cases of disease and the lack of clinical detail on cancer (e.g., stage). Certain national data collection efforts, however, have great potential for cancer-related health services research. AHRQ's HCUP, for example, can be used to assess variations in patterns of care and differences in care across systems of care.

Suggested Citation:"The Data Infrastructure for Health Services Research." Institute of Medicine and National Research Council. 2000. Enhancing Data Systems to Improve the Quality of Cancer Care. Washington, DC: The National Academies Press. doi: 10.17226/9970.
×
Page 37
Suggested Citation:"The Data Infrastructure for Health Services Research." Institute of Medicine and National Research Council. 2000. Enhancing Data Systems to Improve the Quality of Cancer Care. Washington, DC: The National Academies Press. doi: 10.17226/9970.
×
Page 38
Suggested Citation:"The Data Infrastructure for Health Services Research." Institute of Medicine and National Research Council. 2000. Enhancing Data Systems to Improve the Quality of Cancer Care. Washington, DC: The National Academies Press. doi: 10.17226/9970.
×
Page 39
Suggested Citation:"The Data Infrastructure for Health Services Research." Institute of Medicine and National Research Council. 2000. Enhancing Data Systems to Improve the Quality of Cancer Care. Washington, DC: The National Academies Press. doi: 10.17226/9970.
×
Page 40
Suggested Citation:"The Data Infrastructure for Health Services Research." Institute of Medicine and National Research Council. 2000. Enhancing Data Systems to Improve the Quality of Cancer Care. Washington, DC: The National Academies Press. doi: 10.17226/9970.
×
Page 41
Suggested Citation:"The Data Infrastructure for Health Services Research." Institute of Medicine and National Research Council. 2000. Enhancing Data Systems to Improve the Quality of Cancer Care. Washington, DC: The National Academies Press. doi: 10.17226/9970.
×
Page 42
Suggested Citation:"The Data Infrastructure for Health Services Research." Institute of Medicine and National Research Council. 2000. Enhancing Data Systems to Improve the Quality of Cancer Care. Washington, DC: The National Academies Press. doi: 10.17226/9970.
×
Page 43
Suggested Citation:"The Data Infrastructure for Health Services Research." Institute of Medicine and National Research Council. 2000. Enhancing Data Systems to Improve the Quality of Cancer Care. Washington, DC: The National Academies Press. doi: 10.17226/9970.
×
Page 44
Suggested Citation:"The Data Infrastructure for Health Services Research." Institute of Medicine and National Research Council. 2000. Enhancing Data Systems to Improve the Quality of Cancer Care. Washington, DC: The National Academies Press. doi: 10.17226/9970.
×
Page 45
Next: Status of the Cancer Care Data System »
Enhancing Data Systems to Improve the Quality of Cancer Care Get This Book
×
Buy Paperback | $60.00 Buy Ebook | $47.99
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF

One of the barriers to improving the quality of cancer care in the United States is the inadequacy of data systems. Out-of-date or incomplete information about the performance of doctors, hospitals, health plans, and public agencies makes it hard to gauge the quality of care. Augmenting today's data systems could start to fill the gap.

This report examines the strengths and weaknesses of current systems and makes recommendations for enhancing data systems to improve the quality of cancer care. The board's recommendations fall into three key areas:

  • Enhance key elements of the data system infrastructure (i.e., quality-of-care measures, cancer registries and databases, data collection technologies, and analytic capacity).
  • Expand support for analyses of quality of cancer care using existing data systems.
  • Monitor the effectiveness of data systems to promote quality improvement within health systems.
  1. ×

    Welcome to OpenBook!

    You're looking at OpenBook, NAP.edu's online reading room since 1999. Based on feedback from you, our users, we've made some improvements that make it easier than ever to read thousands of publications on our website.

    Do you want to take a quick tour of the OpenBook's features?

    No Thanks Take a Tour »
  2. ×

    Show this book's table of contents, where you can jump to any chapter by name.

    « Back Next »
  3. ×

    ...or use these buttons to go back to the previous chapter or skip to the next one.

    « Back Next »
  4. ×

    Jump up to the previous page or down to the next one. Also, you can type in a page number and press Enter to go directly to that page in the book.

    « Back Next »
  5. ×

    Switch between the Original Pages, where you can read the report as it appeared in print, and Text Pages for the web version, where you can highlight and search the text.

    « Back Next »
  6. ×

    To search the entire text of this book, type in your search term here and press Enter.

    « Back Next »
  7. ×

    Share a link to this book page on your preferred social network or via email.

    « Back Next »
  8. ×

    View our suggested citation for this chapter.

    « Back Next »
  9. ×

    Ready to take your reading offline? Click here to buy this book in print or download it as a free PDF, if available.

    « Back Next »
Stay Connected!