The use of life tables and survival analysis in testing genetic hypotheses, with an application to Alzheimer's disease

Because of the late onset of some neuropsychiatric disorders suspected to be under genetic influence, such as Alzheimer's disease, standard techniques for testing genetic hypotheses are difficult to apply to clinical data. The statistical aspects of life table methods and survival probability estimators which can be used to test such hypotheses have been neglected in the psychiatric literature. Two techniques of this kind, the Weinberg morbidity table and the Kaplan-Meier product limit estimator, are applied to real and simulated data. As estimators of lifetime incidence these methods yield roughly equivalent results for both types of data, although from a theoretical standpoint the original Weinberg estimator appears to suffer from logical defects. Parametric models may offer more definitive results, particularly when an estimator of segregation ratio is required. The clinical data in this report were gathered by interviewing relatives of Alzheimer's disease patients sampled through a nursing home survey in metropolitan Baltimore, Maryland during 1980.

A modified life-table analysis system for cohort studies

A person-years at risk life-table analysis system of computer programs has been developed by the National Institute for Occupational Safety and Health (NIOSH) and is available with detailed documentation. The system was specifically designed to analyze occupational cohort mortality data. These programs require more computer core space and processing time than other available life-table programs. However, the NIOSH programs are advantageous because they include the following: (1) input data editing and modification, (2) mortality rates for 89 cause-of-death categories, (3) assignment of cumulative doses to specific person-years based on either personal or area exposure data, and (4) simultaneous examination of observed and expected deaths by duration of employment (or dose), latency, age, and calendar time.

Assessment of the use of the age- and sex-specific United States population as a control group for analysis of survival in coronary artery disease

The use of the age- and sex-specific U.S. population as a control group for analysis of survival in coronary artery disease was assessed. Population-based survival rates were calculated for nonsurgically treated patients evaluated for coronary artery disease at Duke University Medical Center. Survival of the overall group of medically treated patients with significant coronary artery disease was lower than the corresponding age- and sex-specific U.S. population rates. However, survival of patients with significant disease who had normal left ventricular contraction and stable chest pain was similar to the age- and sex-specific population survival rates. Both the observed survival and the population-based survival estimates for patients with normal left ventricular contraction and stable pain were lower than the survival of patients with normal coronary arteriograms. Even after deaths from ischemic heart disease are eliminated from the population rates, survival of patients with normal coronary arteries exceeded the age-and sex-specific population survival. Because of biases inherent in the selection of patients for cardiac catheterization and the presence of other serious diseases in persons in the general population is not an adequate control group for rigorous analysis of the effect of therapy in coronary artery disease.

Simulation of a health insurance market with adverse selection

A health insurance market is examined in which individuals with a history of high utilization of health care services tend to select fee-for-service (FFS) insurance when offered a choice between FFS and health maintenance organizations (HMOs). In addition, HMOs are assumed to practice community rating of employee groups. Based on these observations and health plan enrollment and premium data from Minneapolis-St. Paul, a deterministic simulation model is constructed to predict equilibrium market shares and premiums for HMO and FFS insurers within a firm. Despite the fact that favorable selection enhances their ability to compete with FFS insurers, the model predicts that HMOs maximize profits at less than 100% market share, and at a lower share than they could conceivably capture. That is, HMOs would not find it to their advantage to drive FFS insurers from the market even if they could. In all cases, however, the profit-maximizing HMO premium is greater than the experience-rated premium and, thus, the average health insurance premium per employee in firms offering both HMOs and FFS insurance is predicted to be greater than in firms offering one experience-rated plan. The model may be used to simulate the effects of varying the employer's method of contributing to health insurance premiums. Several contribution methods are compared. Employers who offer FFS and HMO insurance and pay the full cost of the lowest-cost plan are predicted to have lower average total premiums (employer plus employee contributions) than employers who pay any level percent of the cost of each plan.

A modified life table method to study congenital genetic disorders: an application in sickle cell anemia

A modified life table procedure is introduced designed to study the survival of patients with congenital genetic disorders with the endpoint defined by a complication or death. It uses the ages of the patients as the time axis as in "population' or "current' life tables, and it allows patients to enter and exit the study as in survival life tables. The procedure uses the exact length of time that each patient is observed in the study to determine the conditional probabilities of developing the complication. The proposed procedure is especially helpful in studying recurrent complications or events that occur frequently. The proposed life table procedure is demonstrated in the study of the conditional probability of developing a sickle cell crisis in 509 patients with sickle cell anemia (SS) with different number of prior crises. The demonstration is intended to illustrate the use of the proposed method and not to investigate the clinical severity of sickle cell anemia. It was found that the risk of crisis was to investigate the clinical severity of sickle cell anemia. It was found that the risk of crisis was positively related to the number of prior crises in SS patients (P less than 0.001). This trend was significant in the first three decades of life.

A life table for onset of Huntington's chorea

Accurate quantitative assessment of risk in Huntington's chorea necessitates unbiased estimation of the distribution of age at onset. Previous studies have quoted the mean observed age at onset but this is biased, containing on contribution from heterozygotes who have not manifested, which takes due account of such censored heterozygotes; it demonstrates a considerably older median age onset than hitherto been reported.

The bingo model of survivorship: 1. probabilistic aspects

A "bingo" model is one in which the pattern of survival of a system is determined by whichever of several components, each with its own particular distribution for survival, fails first. The model is motivated by the study of lifespan in animals. A number of properties of such systems are discussed in general. They include the use of a special criterion of skewness that probably corresponds more closely than traditional measures to what the eye observes in casually inspecting data. This criterion is the ratio, r(h), of the probability density at a point an arbitrary distance, h, above the mode to that an equal distance below the mode. If this ratio is positive for all positive arguments, the distribution is considered positively asymmetrical and conversely. Details of the bingo model are worked out for several types of base distributions: the rectangular, the triangular, the logistic, and by numerical methods, the normal, lognormal, and gamma.

Prognostic factors in cancer of the testis: an approach to identification

The clinical experience at the University of Minnesota Hospitals between 1962 and 1977 was reviewed in the hope of defining the biologic features of testicular cancer, knowledge of which is essential to determine the therapeutic needs of each patient. This collaborative effort between clinicians and statisticians has attempted to define important variables in the assessment of risk factors that determine the choice of treatment for testicular cancer. A preliminary analysis of data on the first 78 patients tabulated was performed to test and plan an approach to the final analysis of a larger group of 282 patients. One prognostic factor, staging, is discussed in detail to illustrate the purpose of this exploratory testing. Retrospective analysis, according to stage, of patients who were referred to this cancer center revealed a poorer prognosis than should be anticipated. In reporting the clinical experience of an individual institution, problems of bias are introduced with respect to the nature of the types of tumors studied.

Comparative method for actuarial analysis of cardiac valve replacements

The use of survival analysis in cardiac surgical reports has been widely accepted in order to characterise and compare performances of different valve prostheses. However, the analyses that have been described elsewhere may be improved in several respects. The original method proposed for prosthetic valve evalation examined follow-up related to patients rather than the valves themselves and therefore neglected the possibility of factors other than the valve itself which could affect the fate of cardiac patients. Furthermore, there has not yet been any application of life table methods to compare two different sets of data, or to estimate the average time before an event occurs. The present work defines valve function and malfunction and separates patient survival from valve performance. We define statistical indices, namely the median remaining lifetime and the instantaneous decrement rate. Both indices can be useful; the former measures the average survival time and the latter focuses attention on the intensity of risk changes. Using established statistical theory we calculate the significance level of difference between two independent survival rates. We qualify the presentation of probabilities with a statement of the associated standard errors.

Analysis of life tables with grouping and withdrawals

A number of individuals is observed at the beginning of a period. At the end of the period the number is surviving, the number who have died and the number who have withdrawn are noted. From these three numbers it is required to estimate the death rate for the period. All relevant quantities are supposed independent and identically distributed for the individuals. The likelihood is calculated and found to depend on two parameters, other than the death rate, and to be unidenttifiable so that no consistent estimators exist. For large numbers, the posterior distribution of the death rate is approximated by a normal distribution whose mean is the root of a quadratic equation and whose variance is the sum of two terms; the first is proportional to the reciprocal of the number of individuals, as usually happens with a consistent estimator; the second does not tend to zero and depends on initial opinions about one of the nuisance parameters. The paper is a simple exercise in the routine use of coherent, Bayesian methodology. Numerical calucations illustrate the results.