Actuarial assessment of sex offender recidivism risk: a cross-validation of the RRASOR and the Static-99 in Sweden

We cross-validated two actuarial risk assessment tools, the RRASOR (R. K. Hanson, 1997) and the Static-99 (R. K. Hanson & D. Thornton, 1999), in a retrospective follow-up (mean follow-up time = 3.69 years) of all sex offenders released from Swedish prisons during 1993-1997 (N = 1,400, all men, age > or =18 years). File-based data were collected by a researcher blind to the outcome (registered criminal recidivism), and individual risk factors as well as complete instrument characteristics were explored. Both the RRASOR and the Static-99 showed similar and moderate predictive accuracy for sexual reconvictions whereas the Static-99 exhibited a significantly higher accuracy for the prediction of any violent recidivism as compared to the RRASOR. Although particularly the Static-99 proved moderately robust as an actuarial measure of recidivism risk among sexual offenders in Sweden, both procedures may need further evaluation, for example, with sex offender subpopulations differing ethnically or with respect to offense characteristics. The usefulness of actuarial methods for the assessment of sex offender recidivism risk is discussed in the context of current practice.

Long-term care insurance underwriting: understanding eventual claims experience

Little is known about the accuracy of medical underwriting for long-term care insurance. The lack of data on claims experience continues to be an obstacle in testing the ability of medical underwriting to identify above average financial risks. This study used actual claims data to simulate medical underwriting and to examine the risk, duration, and timing of nursing home use for people with conditions that are uninsurable. The results show that at least one older person in seven who is rejected for long-term care insurance due to underwriting may not represent greater financial risk to insurers than do those who are accepted.

Developing a clinically useful actuarial tool for assessing violence risk

BACKGROUND

A new actuarial method for violence risk assessment--the Iterative Classification Tree (ICT)--has become available. It has a high degree of accuracy but can be time and resource intensive to administer.

AIMS

To increase the clinical utility of the ICT method by restricting the risk factors used to generate the actuarial tool to those commonly available in hospital records or capable of being routinely assessed in clinical practice.

METHOD

A total of 939 male and female civil psychiatric patients between 18 and 40 years old were assessed on 106 risk factors in the hospital and monitored for violence to others during the first 20 weeks after discharge.

RESULTS

The ICT classified 72.6% of the sample as either low risk (less than half of the sample's base rate of violence) or high risk (more than twice the sample's base rate of violence).

CONCLUSIONS

A clinically useful actuarial method exists to assist in violence risk assessment.

Parametric cure models of relative and cause-specific survival for grouped survival times

With parametric cure models, we can express survival parameters (e.g. cured fraction, location and scale parameters) as functions of covariates. These models can measure survival from a specific disease process, either by examining deaths due to the cause under study (cause-specific survival), or by comparing all deaths to those in a matched control population (relative survival). We present a binomial maximum likelihood algorithm to be used for actuarial data, where follow-up times are grouped into specific intervals. Our algorithm provides simultaneous maximum likelihood estimates for all the parameters of a cure model and can be used for cause-specific or relative survival analysis with a variety of survival distributions. Current software does not provide the flexibility of this unified approach.

Improving risk assessments for sex offenders: a comparison of three actuarial scales

The study compared the predictive accuracy of three sex offender risk-assessment measures: the RRASOR (Hanson, 1997), Thornton's SACJ-Min (Grubin, 1998), and a new scale, Static-99, created by combining the items from the RRASOR and SACJ-Min. Predictive accuracy was tested using four diverse datasets drawn from Canada and the United Kingdom (total n = 1301). The RRASOR and the SACJ-Min showed roughly equivalent predictive accuracy, and the combination of the two scales was more accurate than either original scale. Static-99 showed moderate predictive accuracy for both sexual recidivism (r = 0.33, ROC area = 0.71) and violent (including sexual) recidivism (r = 0.32, ROC area = 0.69). The variation in the predictive accuracy of Static-99 across the four samples was no more than would be expected by chance.

A classification tree approach to the development of actuarial violence risk assessment tools

Since the 1970s, a wide body of research has suggested that the accuracy of clinical risk assessments of violence might be increased if clinicians used actuarial tools. Despite considerable progress in recent years in the development of such tools for violence risk assessment, they remain primarily research instruments, largely ignored in daily clinical practice. We argue that because most existing actuarial tools are based on a main effects regression approach, they do not adequately reflect the contingent nature of the clinical assessment processes. To enhance the use of actuarial violence risk assessment tools, we propose a classification tree rather than a main effects regression approach. In addition, we suggest that by employing two decision thresholds for identifying high- and low-risk cases--instead of the standard single threshold--the use of actuarial tools to make dichotomous risk classification decisions may be further enhanced. These claims are supported with empirical data from the MacArthur Violence Risk Assessment Study.

End points in the analysis of breast cancer survival: relapse versus death from tumor

BACKGROUND

To determine whether relapse and death from tumor are comparable as survival end points for assessing therapeutic efficacy, five prospective, randomized clinical trials of adjuvant therapy for stage II breast cancer were analyzed. One thousand eight hundred ninety patients were combined from five clinical groups into a single group for analysis.

METHODS

Actuarial and parametric survival methods were used to generate three estimates for the likelihood of cure (LOC): (1) for all patients, with relapse as the end point to survival (LOCR); (2) for all patients, with death from tumor as the end point (LOCD); and (3) for patients with relapse only with death from tumor as the end point (LOCRD). Linear regression analysis was used to compare time to relapse for each patient with time from relapse to death.

RESULTS

Estimates of LOCR ranged from 33.5% to 38.4%, estimates of LOCD ranged from 36.3% to 44.2%, and estimates of LOCRD ranged from 0% to 6%. Thus LOCR and LOCD are approximately equal for these patients. On the other hand, time to relapse correlated poorly with time from relapse to death (r2 = 0.005).

CONCLUSIONS

If therapy affects LOC, relapse and death should ultimately lead to the same conclusion with respect to therapeutic efficacy, because both end points lead to essentially the same LOC. If therapy affects time to relapse, however, these two end points may ultimately lead to different conclusions, because time to relapse correlates poorly with time from relapse to death.

Increasing prostate-specific antigen profile following definitive radiation therapy for localized prostate cancer: clinical observations

PURPOSE

To examine the natural history of patients who have received definitive radiation therapy alone for clinically localized prostate cancer and have an increasing prostate-specific antigen (PSA) profile.

PATIENTS AND METHODS

One hundred fifty-one men with an increasing PSA profile after definitive radiotherapy were identified. The subsequent natural history of these men, including local recurrence, distant metastasis, and survival, was examined. In 119 men, posttreatment PSA doubling times (PSADT) were calculated using linear regression. Cox regression models were used to examine the effect of clinical and treatment variables on clinical failure and survival.

RESULTS

Patients with high pretreatment PSA values, high Gleason scores, and T3 tumors were more likely to develop a PSA elevation. The median calculated post-treatment PSADT was 13 months, and 95% of patients had posttreatment PSADT of less than 3 years. PSADT was correlated with tumor stage and Gleason score. Five years after PSA elevation, the estimated rate of clinical local recurrence is 26% and the estimated rate of distant metastases is 47%. Rapid PSADT (< 12 months) and a short interval from the end of treatment to PSA elevation (< 12 months) were significant independent predictors of distant metastases. The estimated rates of overall and cause-specific survival 5 years after PSA elevation are 65% and 76%, respectively. Gleason grade is the only significant independent predictor of overall and cause-specific survival after PSA elevation.

CONCLUSION

The natural history of men who have an increasing PSA profile following definitive radiotherapy is heterogeneous. In the absence of salvage therapy, at least three quarters of men will have clinical evidence of recurrent disease 5 years after a PSA elevation is detected. Men with a rapid posttreatment PSADT and a short interval from the end of treatment to an increasing PSA profile are at a very high risk of developing distant metastasis within 5 years of PSA elevation.

Using actuarial models to assess managed care risk

Actuarial models can be used to illustrate the financial risks inherent in healthcare provider operations. They are especially useful in analyzing the implications of managed care agreements, which revolve around risk and the financing of risk. Simple actuarial models may focus on inpatient utilization and reimbursement under capitation. More sophisticated models detail risks associated with individual diagnosis-related groups, as well as with many types of outpatient and physician services. Actuarial models can provide an objective basis for planning for the future and can be used to build consensus on strategies that will ensure success under managed care agreements.

[The characteristics of actuarial calculations for voluntary medical insurance]

Presents methods of mathematical validation of agreements on voluntary medical insurance. Reduction coefficients are used for determining the onset of the insurer's responsibility after a certain number of invalidity days. The reduction coefficient is regarded as a probability that the insured person recovers not later than by a certain date, and the sequence of these coefficients as function of distribution of the course of invalidity. Use of the actuarial estimations for validating different aspects of voluntary medical insurance helps specify the probable payments and provide the financial stability of insurance companies.

Risk adjusting community rated health plan premiums: a survey of risk assessment literature and policy applications

This paper surveys recent health care reform debates and empirical evidence regarding the potential role for risk adjusters in addressing the problem of competitive risk segmentation under capitated financing. We discuss features of health plan markets affecting risk selection, methodological considerations in measuring it, and alternative approaches to financial correction for risk differentials. The appropriate approach to assessing risk differences between health plans depends upon the nature of market risk selection allowed under a given reform scenario. Because per capita costs depend on a health plan's population risk, efficiency, and quality of service, risk adjustment will most strongly promote efficiency in environments with commensurately strong incentives for quality care.

Pitfalls of inferring annual mortality from inspection of published survival curves

In many FU articles currently published, results are given primarily in the form of graphs of survival curves, rather than in the form of life table data. Sometimes the authors may comment on the slope of the survival curve as though it were equal to the annual mortality rate (after reversal of the minus sign to a plus sign). Even if no comment of this sort is made, medical directors and underwriters may be tempted to think along similar lines in trying to interpret the significance of the survival curve in terms of mortality. However it is a very serious error of life table methodology to conceive of mortality rate as equal to the negative slope of the survival curve. The nature of the error is demonstrated in this article. An annual mortality rate derived from the survival curve actually depends on two variables: a quotient with the negative slope (sign reversed), delta P/ delta as the numerator, and the survival rate, P, itself as the denominator. The implications of this relationship are discussed. If there are two "parallel" survival curves with the same slope at a given time duration, the lower curve will have a higher mortality rate than the upper curve. A constant slope with increasing duration means that the annual mortality rate also increases with duration. Some characteristics of high initial mortality are also discussed and their relation to different units of FU time.(ABSTRACT TRUNCATED AT 250 WORDS)

Improved methodology for analyzing local and distant recurrence

Studies of radiation therapy and/or surgery in the treatment of cancer frequently use actuarial methods to estimate curves of time to local failure and compare two such curves with statistical methods originally developed for survival data. In such analyses, patients who fail first in distant sites or die before local failure are considered censored for time to local failure. While the arithmetic of these calculations is usually correct, the interpretation of the results is almost universally incorrect. For example, an actuarial Kaplan-Meier curve of time to breast recurrence after breast conserving treatment consistently overestimates the percentage of patients who would benefit from a subsequent mastectomy. Actuarial methods require the assumption that time to local failure and time to distant failure are statistically independent. For most human malignancies this is not a reasonable assumption, since there are always some patient subgroups at high risk of both local and distant failure and some patient subgroups at low risk for either type of failure. Without the assumption of independence, the time to local failure distribution is not well defined. The basic problem is that estimating time to local failure falls into the category of analyzing "competing risks," since the various causes of failure are competing for the same patient. For this reason, the effects of a particular treatment on local failure cannot be assessed separately from its effects on distant failure. This report explains the concepts of statistical independence, nonidentifiability, and competing risks and illustrates the pitfalls of using actuarial methods to assess local tumor control.(ABSTRACT TRUNCATED AT 250 WORDS)

Life-span of amalgam restorations in primary molars: some results and comments on statistical analyses

The life-span of amalgam restorations placed in primary molars by three general dental practitioners is investigated. Two methods of analysis have been carried out. In the analysis of "non-independent" restorations (i.e. using several restorations from each mouth), the median survival time was 52.8 months with an apparent standard error of 2.1 months. In the analysis of "independent" restorations (i.e. only using one restoration from each mouth), three random samples were chosen and their median survival times were 68.2 months (s.e. = 6.9), 60.5 months (s.e. = 6.9), and 56.8 months (s.e. = 3.7). The two methods are compared and discussed. It is concluded that the analysis of "independent" restorations should be the method of choice in studying the life-span of restorations.

Physicians who have lost their malpractice insurance. Their demographic characteristics and the surplus-lines companies that insure them

The present study analyzes demographic data on 920 physicians who lost their coverage and applied to a "surplus-lines" company that insures essentially all applicants. Our analysis reveals that (1) some specialties are heavily overrepresented in the surplus-lines pool, (2) physicians aged 45 to 54 years are also overrepresented, (3) board certification is seen as frequently in the surplus-lines group as in the US physician population, and (4) the percentage of foreign medical graduates in the surplus-lines pool is virtually the same as that in the US physician population. A model of the actuarial process by which claims data can lead to termination of standard coverage suggests that disproportionate representation of high-risk specialties is not simply a function of a high average claims rate. We also show that, in contrast to joint underwriting associations, surplus-lines companies impose high premiums, large deductibles, and restrictions on practice, all of which are likely to reduce the frequency of negligent behavior.