A comparison of physical therapy, chiropractic manipulation, and provision of an educational booklet for the treatment of patients with low back pain

BACKGROUND AND METHODS

There are few data on the relative effectiveness and costs of treatments for low back pain. We randomly assigned 321 adults with low back pain that persisted for seven days after a primary care visit to the McKenzie method of physical therapy, chiropractic manipulation, or a minimal intervention (provision of an educational booklet). Patients with sciatica were excluded. Physical therapy or chiropractic manipulation was provided for one month (the number of visits was determined by the practitioner but was limited to a maximum of nine); patients were followed for a total of two years. The bothersomeness of symptoms was measured on an 11-point scale, and the level of dysfunction was measured on the 24-point Roland Disability Scale.

RESULTS

After adjustment for base-line differences, the chiropractic group had less severe symptoms than the booklet group at four weeks (P=0.02), and there was a trend toward less severe symptoms in the physical therapy group (P=0.06). However, these differences were small and not significant after transformations of the data to adjust for their non-normal distribution. Differences in the extent of dysfunction among the groups were small and approached significance only at one year, with greater dysfunction in the booklet group than in the other two groups (P=0.05). For all outcomes, there were no significant differences between the physical-therapy and chiropractic groups and no significant differences among the groups in the numbers of days of reduced activity or missed work or in recurrences of back pain. About 75 percent of the subjects in the therapy groups rated their care as very good or excellent, as compared with about 30 percent of the subjects in the booklet group (P<0.001). Over a two-year period, the mean costs of care were $437 for the physical-therapy group, $429 for the chiropractic group, and $153 for the booklet group.

CONCLUSIONS

For patients with low back pain, the McKenzie method of physical therapy and chiropractic manipulation had similar effects and costs, and patients receiving these treatments had only marginally better outcomes than those receiving the minimal intervention of an educational booklet. Whether the limited benefits of these treatments are worth the additional costs is open to question.

Medication use for low back pain in primary care

STUDY DESIGN

A longitudinal observational study of primary care patients with low back pain.

OBJECTIVES

1) To describe medications prescribed for back pain, 2) to identify patient characteristics associated with type of drug therapy, 3) to determine if the prescription of certain drugs is associated with better outcomes, and 4) to compare physician prescribing behavior with national guidelines.

SUMMARY OF BACKGROUND DATA

Few previous studies have focused on medication prescribing patterns for back pain in primary care.

METHODS

Two-hundred nineteen patients aged 20-69 years who were making a first visit for an episode of back pain were studied. After the visit, patients completed questionnaires regarding sociodemographic characteristics, health status, back pain experience, and use of medications. Symptom severity and dysfunction were assessed by telephone 1 week after the visit.

RESULTS

Sixty-nine percent of patients were prescribed nonsteroidal anti-inflammatory drugs, 35% muscle relaxants, 12% narcotics, and 4% acetaminophen. Twenty percent received no medications. Patients were more likely to receive medications if they had a desire for medication, pain below the knee, less than 3 weeks of pain before visit, more severe symptoms, or greater dysfunction. Patients with more severe symptoms were more likely to receive narcotics or muscle relaxants. Patients with greater dysfunction were also more likely to receive narcotics. Type of drug therapy predicted symptom severity but not dysfunction after 1 week. Controlling for other factors, those receiving medications had less severe symptoms after 1 week than patients who received no medication. Patients receiving both muscle relaxants and nonsteroidal anti-inflammatory drugs had the best outcomes. Medication use for back pain in this health maintenance organization was generally concordant with national guidelines.

CONCLUSIONS

Nonsteroidal anti-inflammatory drugs, often augmented by muscle relaxants, are a standard medical treatment for back pain in primary care. In this observational study, patients prescribed medications, particularly muscle relaxants, reported less severe symptoms after 1 week than those receiving no medications. However, randomized trials are needed to determine which medication or combinations of medications are most effective.

Breast cancer survival and treatment in health maintenance organization and fee-for-service settings

BACKGROUND

Enrollment in health maintenance organizations (HMOs) has increased rapidly during the past 10 years, reflecting a growing emphasis on health care cost containment. To determine whether there is a difference in the treatment and outcome for female patients with breast cancer enrolled in HMOs versus a fee-for-service setting, we compared the 10-year survival and initial treatment of patients with breast cancer enrolled in both types of plans.

METHODS

With the use of tumor registries covering the greater San Francisco-Oakland and Seattle-Puget Sound areas, respectively, we obtained information on the treatment and outcome for 13,358 female patients with breast cancer, aged 65 years and older, diagnosed between 1985 and 1992. We linked registry information with Medicare data and data from the two large HMOs included in the study. We compared the survival and treatment differences between HMO and fee-for-service care after adjusting for tumor stage, comorbidity, and sociodemographic characteristics.

RESULTS

In San Francisco-Oakland, the 10-year adjusted risk ratio for breast cancer deaths among HMO patients compared with fee-for-service patients was 0.71 (95% confidence interval [CI] = 0.59-0.87) and was comparable for all deaths. In Seattle-Puget Sound, the risk ratio for breast cancer deaths was 1.01 (95% CI = 0.77-1.33) but somewhat lower for all deaths. Women enrolled in HMOs were more likely to receive breast-conserving surgery than women in fee-for-service (odds ratio = 1.55 in San Francisco-Oakland; 3.39 in Seattle). HMO enrollees undergoing breast-conserving surgery were also more likely to receive adjuvant radiotherapy (San Francisco-Oakland odds ratio = 2.49; Seattle odds ratio = 4.62).

CONCLUSIONS

Long-term survival outcomes in the two prepaid group practice HMOs in this study were at least equal to, and possibly better than, outcomes in the fee-for-service system. In addition, the use of recommended therapy for early stage breast cancer was more frequent in the two HMOs.

Use of exogenous hormones and risk of temporomandibular disorder pain

Temporomandibular disorders (TMD) are common pain conditions that have their highest prevalence among women of reproductive age. The higher prevalence of TMD pain among women, pattern of onset after puberty and lowered prevalence rates in the post-menopausal years suggest that female reproductive hormones may play an etiologic role in TMD. Two epidemiologic studies were designed to assess whether use of exogenous hormones is associated with increased risk of TMD pain. Both used data from automated pharmacy records of women enrolled in a large health maintenance organization to identify prescriptions filled for post-menopausal hormone replacement therapies (Study 1) or for oral contraceptives (OCs) (Study 2). Study 1 employed an age-matched case-control design to compare post-menopausal hormone use among 1291 women over age 40 referred for TMD treatment and 5164 controls not referred. After controlling for health services use, the odds of being a TMD case were approximately 30% higher among those receiving estrogen compared to those not exposed (P = 0.002); a clear dose-response relationship was evident. The relationship of progestin use to TMD was not statistically significant. Study 2 used a similar design to examine the relationship of OC use to referral for TMD care, drawing on data from 1473 cases and 5892 controls aged 15-35. Use of OCs was also associated with referral for TMD care, with an increased risk of TMD of approximately 20% for OC users, after controlling for health services use (P < 0.05). These results suggest that female reproductive hormones may play an etiologic role in orofacial pain. This relationship warrants further investigation through epidemiologic, clinical and basic research.

Predicting poor outcomes for back pain seen in primary care using patients' own criteria

STUDY DESIGN

A prospective cohort study of patients seen in primary care for low back pain.

OBJECTIVES

A new measure of back pain outcomes is used to describe the status of back problems at various intervals after visits to primary care physicians and to identify subsets of patients with worse prognoses.

SUMMARY OF BACKGROUND DATA

Most previous studies of the prognosis of back pain in primary care have failed to provide clinically useful information.

METHODS

Baseline data were collected from 219 patients making an initial visit for an episode of low back pain to a primary care clinic. A measure of how patients reported they would feel if they had their current back symptoms for the rest of their lives ("Symptom Satisfaction") was used to distinguish good from poor outcomes. Patient outcomes were assessed 1, 3, 7, and 52 weeks after the index visit.

RESULTS

Only 67% of patients reported good outcomes after 7 weeks, and only 71% were satisfied with their condition 1 year later. After controlling for the effects of other variables measured during the initial physician visit, only younger age, depression, and pain below the knee were significant predictors of poor outcome at 7 weeks, and only pain below the knee and depression were significant predictors at 1 year.

CONCLUSIONS

The proportion of primary care patients with back pain who have poor outcomes appears to be higher than generally recognized. Ways of improving how primary care responds to patients with persisting pain should be investigated.

A prospective study of contact lens complications in a managed care setting

BACKGROUND

Understanding the rate of contact lensrelated complications and the factors that affect their occurrence can facilitate better prescribing decisions.

METHODS

In a managed care setting, 1496 patient visits were evaluated using a common protocol by 11 optometrists to determine the prevalence of all contact lens-related complications.

RESULTS

Over one-half (61%) of the visits were normal, with the remainder showing some type of complication. The more prevalent complications included superficial punctate staining (17.3%) and neovascularization (11.4%). Total complications were less prevalent with rigid gas permeable (RGP) and disposable lens types. Planned replacement soft lenses, used on a daily wear schedule, had the lowest prevalence of more serious complications when compared to conventional soft and disposable lenses. Patients on an extended wear schedule greater than 3 days were more likely to experience complications. The use of nonapproved care systems showed more complications, with serious complications reduced when a one-step care system was used.

CONCLUSION

Choice of lens type, wearing schedule, and care system does affect the prevalence of complications, which underlines the importance of the recommendations of the prescriber.

Predictors of chronic benzodiazepine use in a health maintenance organization sample

While expert recommendations caution against long-term benzodiazepine use in the elderly, survey data suggest increasing benzodiazepine use with age. Computerized pharmacy records of staff-model HMO were used to examine benzodiazepine prescribing. Six-month prevalence of benzodiazepine use (2.8%) and prevalence of continued use (0.7%) were lower than earlier reports. Prevalence was higher in women and increased steadily with age. Among 7012 patients beginning benzodiazepine treatment, duration of use increased with patient age, prescription by a psychiatrist (vs. primary care or medical/surgical specialist), use of higher-potency drugs (lorazepam, and alprazolem, clonazepam) and larger number of pills in the initial prescription. Individual physicians varied significantly in drug choice, initial prescription size, and likelihood of chronic use. Among 200 patients treated in primary care, the physician-recorded indication for prescription was anxiety or depression in 27%, insomnia in 20%, and pain symptoms in 38%. These findings indicate a gap between benzodiazepine efficacy research and current clinical practice.

Measurement of interrater agreement with adjustment for covariates

The kappa coefficient measures chance-corrected agreement between two observers in the dichotomous classification of subjects. The marginal probability of classification by each rater may depend on one or more confounding variables, however. Failure to account for these confounders may lead to inflated estimates of agreement. A multinomial model is used that assumes both raters have the same marginal probability of classification, but this probability may depend on one or more covariates. The model may be fit using software for conditional logistic regression. Additionally, likelihood-based confidence intervals for the parameter representing agreement may be computed. A simple example is discussed to illustrate model-fitting and application of the technique.

Pitfalls of patient education. Limited success of a program for back pain in primary care

STUDY DESIGN

Low back pain patients seen in primary care were allocated randomly to one of two educational interventions or to usual care.

OBJECTIVE

To evaluate educational interventions designed to improve the outcomes of primary care for low back pain.

SUMMARY OF BACKGROUND DATA

Patients with back pain are frequently dissatisfied with their medical care and identify lack of information as the most insufficient aspect.

METHODS

In a large Health Maintenance Organization clinic, 293 subjects were allocated randomly to receive usual care, an educational booklet, or a 15-minute session with a clinic nurse, including the booklet and a follow-up telephone call. Outcome measures included satisfaction with care, perceived knowledge, participation in exercise, functional status, symptom relief, and health care use. Outcomes were assessed 1, 3, 7, and 52 weeks after the intervention.

RESULTS

The nurse intervention resulted in higher patient satisfaction than usual care (P < 0.001) and higher perceived knowledge (P < 0.001). Self-reported exercise participation was also higher in the nurse intervention group after a 1-week follow-up period (97% vs. 65% in the other groups; P < 0.0001). There were no significant differences among the three groups in worry, symptoms, functional status, or health care use at any follow-up interval. Differences in self-reported exercise and perceived knowledge were no longer significant after 7 weeks.

CONCLUSIONS

These findings challenge the value of purely educational approaches in reducing functional impact or health care use related to back pain and also challenge the value of fitness exercise in the most acute phase of back pain.

Health care costs of primary care patients with recognized depression

BACKGROUND

While an extensive literature documents the influence of depression on general medical services utilization, estimates of the economic burden of depression have focused on the direct costs of depression treatment. Higher use of general medical services may contribute significantly to the true cost of depressive illness.

METHODS

Computerized record systems of a large staff-model health maintenance organization (HMO) were used to identify consecutive primary care patients with visit diagnoses of depression (n = 6257) and a comparison sample of primary care patients with no depression diagnosis (n = 6257). The HMO accounting records were used to compare components of health care costs.

RESULTS

Patients diagnosed as depressed had higher annual health care costs ($4246 vs $2371, P < .001) and higher costs for every category of care (eg, primary care, medical specialty, medical inpatient, pharmacy, laboratory). Similar cost differences were observed for each of the subgroups examined (patients treated with antidepressants, those not treated with antidepressants, and those diagnosed at routine physical examination visits). Pharmacy records indicated greater chronic medical illness in the diagnosed depression group, but large cost differences remained after adjustment ($3971 vs $2644). Twofold cost differences persisted for at least 12 months after initiation of treatment.

CONCLUSIONS

Diagnosis of depression is associated with a generalized increase in use of health services that is only partially explained by comorbid medical conditions. In the primary care sector, this greater medical utilization exceeds direct treatment costs for depression. The persistence of utilization differences suggests that recognition and initiation of treatment alone are not adequate to reduce utilization differences.

Stage, age, comorbidity, and direct costs of colon, prostate, and breast cancer care

PURPOSE

This study was conducted to evaluate the effect of stage at diagnosis, age, and level of comorbidity (presence of other illness) on the costs of treating three types of cancer among members of a health maintenance organization.

METHODS

Among 388,000 members enrolled anytime during 1990 and 1991 in Group Health Cooperative (GHC) of Puget Sound (Washington State), we estimated the total and net direct costs of medical care for colon, prostate, and breast cancers, including both incident (290, 554, and 645 patients, respectively) and prevalent (1046, 1295, and 2299 patients, respectively) cases. We summarized costs for initial, continuing, and terminal phases of care. Net costs were the difference between the costs of the care of each case subject and the average costs of the care for all enrollees without the cancer of interest who were of the same sex and in the same 5-year age group. Differences in estimated total and net costs by stage at diagnosis, age, and comorbidity were separately evaluated using multivariate regression modeling. All P values were two-sided. Comorbidity was based on a score calculated from 1988 pharmacy data.

RESULTS

Total costs of initial care increased with stage at diagnosis for colon (P = .0013) and breast (P < .0001) cancer cases, but not for prostate cancer cases. Total initial costs decreased with age for prostate (P = .0225) and breast (P = .0002) cancers but did not change with degree of comorbidity for any of the three cancers. Total continuing medical care costs increased with stage at diagnosis for colon (P < .0001) and breast (P < .0001) cancer cases but not for prostate cancer cases. Total terminal care costs were similar by stage for all three cancers. Net initial costs differed with stage for all three cancers (P < .05). Net continuing care costs increased with stage (P < .0001) and decreased with age (P < .001) for colon and breast cancers but not for prostate cancer. Net continuing care costs decreased with comorbidity for all three cancers (P = .004, P = .011, and P < .0001 for colon, prostate, and breast cancers, respectively). Among regional stage cancers, continuing care costs decreased with age for colon (P < .0017) and breast (P = .033) cancers but not for prostate cancers.

CONCLUSIONS

The results show that total costs vary by stage at diagnosis and age, but the patterns of variation differ for each cancer. Costs of cancer are not simply additive to costs of other conditions.

IMPLICATIONS

More needs to be done to explore the reasons and implications of age-related cost differences. Cost-effectiveness analyses of cancer control interventions that shift cancer stage distributions may need to consider both the age and comorbidity of the target populations.

Health care costs associated with depressive and anxiety disorders in primary care

OBJECTIVE

The authors examined the overall health care costs associated with depression and anxiety among primary care patients.

METHOD

Of 2,110 consecutive primary care patients in a health maintenance organization, 1,962 were screened with the 12-item General Health Questionnaire. A stratified random sample of 615 patients were selected for further diagnostic assessment; 373 of these patients completed the Composite International Diagnostic Interview at baseline and 328 were reassessed 12 months later. Computerized cost records were used to calculate total health care costs for the 6-month period surrounding the baseline assessment and a similar period surrounding the follow-up assessment. Cost accounting data were available for 327 patients at baseline and for 206 patients at both assessments.

RESULTS

Primary care patients with DSM-III-R anxiety or depressive disorders at baseline had markedly higher baseline costs ($2,390) than patients with subthreshold disorders ($1,098) and those with no anxiety or depressive disorder ($1,397). Large cost differences persisted after adjustment for medical morbidity. Cost differences reflected higher utilization of general medical services rather than higher mental health treatment costs. Although most patients with baseline anxiety or depressive disorders showed significant improvement, longitudinal analyses did not show any clear relationship between change in psychiatric diagnosis and change in health care cost.

CONCLUSIONS

Among primary care patients, anxiety and depressive disorders are associated with markedly higher health care costs even after adjustment for medical comorbidity. In this small sample, improvement in depression over 1 year was not clearly associated with decreases in cost.

Effects of practice style in managing back pain

OBJECTIVE

To assess the effects of a practice style of back pain management consistent with self-care (infrequent prescribing of pain medications and bed rest) on long-term functional outcomes, costs of care, and patient satisfaction.

DESIGN

A quasi-experimental observational study in which primary care physicians (n = 44) were categorized according to one of three practice style groups defined by a low, moderate, or high frequency of prescribing pain medications and bed rest for many patients (average, 24 patients per physician).

SETTING

Primary care practices of a large, staff model health maintenance organization, Group Health Cooperative of Puget Sound.

PATIENTS

Consecutive patients with back or neck pain of participating primary care physicians. Patients were interviewed 1 month (n = 1071) and 1 year and 2 years (n = 911) after their index visits.

RESULTS

Patients in the three practice style groups rated similarly the quality of medical care received for back pain. Patients treated by physicians who infrequently prescribed pain medications and bed rest were more satisfied with education about back pain. On a scale of 0 to 10, the mean rating of agreement with the statement, "After your visit the doctor, you fully understood how to take care of your back problem," was 5.6 +/- 3.6 among patients of physicians who frequently prescribed medication and rest and was 6.6 +/- 3.5) among those who infrequently prescribed medication and bed rest. At 1 month, 30% of patients of physicians who infrequently prescribed medications and bed rest were graded as having moderate to severe activity limitation because of back pain, whereas 37% of patients in the moderate group had this grading, and 46% of patients of physicians who frequently prescribed were graded as having moderate to severe activity limitation. Differences in activity limitation by practice style group were no longer evident at 1 or 2 years of follow-up. The total 1-year costs of back care were higher among patients seen by physicians who frequently prescribed bed rest and pain medications (cost, $768 +/- $1592) than among those seen by physicians who infrequently prescribed (cost, $428 +/- $665), due largely to differences in inpatient and specialty care costs. The adjusted difference in costs, after controlling for case-mix variables, was $277 (95% Cl, $85.50 to $471.32).

CONCLUSIONS

A practice style consistent with back pain self-care yielded similar long-term pain and functional outcomes at lower cost and was associated with higher satisfaction with patient education compared with a practice style characterized by more frequent prescribing of pain medications and bed rest.

Patterns of antidepressant use in community practice

Computerized pharmacy records from a large staff-model health maintenance organization were used to examine patterns of antidepressant use by primary care physicians and psychiatrists. Based on timing of prescription refills, patients treated by psychiatrists were more likely than those treated in primary care to continue medication for more than 30 days (35% vs 25%, p < 0.00001) and more likely to reach a prescribed daily dose of 100 mg of imipramine or the equivalent (48% vs 40%, p < 0.00001). Patients treated with newer antidepressants were significantly more likely to continue treatment past 30 days (range from 75% for fluoxetine to 54% for doxepin, p < 0.00001) and to reach an adequate daily dose (range from 51% for fluoxetine to 26% for doxepin, p < 0.00001). Psychiatrists more often prescribed newer antidepressants, and much of the difference between specialties could be explained by drug selection. These findings suggest more intensive antidepressant treatment than in earlier reports, especially in primary care. More intensive treatment with newer antidepressants may reflect more tolerable side effects, but these observational data are liable to selection bias. Any potential advantages of newer antidepressant medications must be balanced against significantly higher costs.

Back pain in primary care. Outcomes at 1 year

Outcomes of primary care back pain patients (N = 1128) were studied at 1 year after seeking care. Changes in depression depending on outcome, and predictors of poor outcome were evaluated. Less than one back pain patient in five reported recent onset (first onset within the previous 6 months). One year after seeking care, the large majority of both recent and nonrecent-onset patients reported having back pain in the previous month (69% vs. 82%). A significant minority of both recent and nonrecent-onset patients had either a poor functional outcome (14% vs. 21%) or continuing high intensity pain without appreciable disability (10% vs. 16%). Predictors of poor outcome included pain-related disability, days in pain, lower educational attainment, and female gender. Among initially dysfunctional patients with persistent pain, one half were improved and one third had a good outcome at the 1-year follow-up. Among initially dysfunctional patients who experienced a good outcome, elevated depressive symptoms improved to normal levels at follow-up. The outcome of back pain was predicted by pain-related disability and days in pain rather than by recency of onset, so it may be more meaningful to distinguish characteristic levels of pain intensity, pain-related disability, and pain persistence than to classify patients as acute or chronic.

The impact of physician attitudes on patient satisfaction with care for low back pain

We wished to determine whether patient satisfaction was related to physicians' confidence in their abilities to effectively manage low back pain, and to examine their attitudes about patients with back pain. The confidence and attitudes of primary care providers were determined using self-administered questionnaires. Patient satisfaction with care was assessed during telephone interviews conducted 3 weeks after a clinic visit for low back pain. The study was conducted in a primary care clinic of a large health maintenance organization. Completed surveys were obtained from 21 primary care providers (18 physicians and three physician assistants) and 270 of their patients with low back pain. Three satisfaction scales specific to low back pain were used to measure patient satisfaction with regard to information received from provider, caring, and effectiveness of treatment. The results showed that the providers' attitudes about patients with low back pain were not associated with any of the patient satisfaction measures. However, patients of more confident providers were significantly more satisfied with the information they received than were patients of less confident providers. These differences could not be explained by years in practice, length of visit, patient demographics, or the severity and duration of low back pain. These findings suggest that providers who have more confidence in their abilities to effectively manage low back pain may in fact be more effective patient educators.

Examination of the patient with a knee dislocation. The case for selective arteriography

One hundred fifteen patients with a unilateral knee dislocation underwent arteriography to examine the popliteal artery. The incidence of popliteal artery injury was 23% (27 patients). Clinically, 29 (25%) of the 115 patients had an abnormal ipsilateral pedal pulse and 23 (79%) of these 29 patients had an arteriographically identified popliteal artery injury. Twenty-two arteries were surgically repaired and one was treated without surgery. Eight-six patients had normal pulses; the arteriogram showed no abnormalities in 77, demonstrated spasm in five, and revealed an intimal flap in four. All 86 patients were treated without surgery and had no delayed vascular complications. This demonstrates that the vascular examination is an accurate predictor of major popliteal artery injury following knee dislocation. Patients with an abnormal pedal pulse warrant arteriography due to a high incidence (79%) of popliteal artery injury. Patients with normal pulses may be monitored by clinical examination only. Popliteal artery injuries in this group are minor and rarely require intervention.

The evolving surgical management of recurrent carotid stenosis

The traditional approach to recurrent carotid stenosis has been repeat endarterectomy or patch angioplasty. Concern with the durability of repeat carotid endarterectomy has resulted in our use of carotid resection with autogenous graft interposition. This study was designed to determine the outcome and efficacy of carotid resection compared with repeat carotid endarterectomy in the management of recurrent carotid stenosis. From 1974 to 1991, 162 operations (repeat carotid endarterectomy 105, carotid resection 57) were performed for recurrent carotid stenosis. Indication for operation was hemispheric symptoms in 63% of patients, nonlateralizing symptoms in 25%, asymptomatic stenosis in 7%, and previous stroke in 5%. Ninety-one percent of patients had stenosis greater than 90% on arteriography. The perioperative stroke rate for carotid resection was 3.5%, with a subsequent rate of 0.0064 strokes per year. For repeat carotid endarterectomy, the perioperative stroke rate was 1.9% with a subsequent rate of 0.011 strokes per year. Graft patency after carotid resection was 93% (mean follow-up, 35 months). Four patients treated with carotid resection had graft thrombosis, and two of the four remained asymptomatic. After repeat carotid endarterectomy, one patient had carotid thrombosis, and recurrent stenosis greater than 50% developed in 23 patients (mean follow-up, 64 months). Twenty patients treated with repeat carotid endarterectomy underwent an additional operation for further symptomatic recurrent carotid stenosis. We conclude carotid resection is a safe and effective alternative to repeat carotid endarterectomy for patients undergoing operation for recurrent carotid stenosis.

A comparison of methods for calculating a stratified kappa

Investigators use the kappa coefficient to measure chance-corrected agreement among observers in the classification of subjects into nominal categories. The marginal probability of classification may depend, however, on one or more confounding variables. We consider assessment of interrater agreement with subjects grouped into strata on the basis of these confounders. We assume overall agreement across strata is constant and consider a stratified index of agreement, or 'stratified kappa', based on weighted summations of the individual kappas. We use three weighting schemes: (1) equal weighting; (2) weighting by the size of the table; and (3) weighting by the inverse of the variance. In a simulation study we compare these methods under differing probability structures and differing sample sizes for the tables. We find weighting by sample size moderately efficient under most conditions. We illustrate the techniques by assessing agreement between surgeons and graders of fundus photographs with respect to retinal characteristics, with stratification by initial severity of the disease.

Methods, statistical features, and baseline results of a standardized, multicentered ophthalmologic surgical trial: the Silicone Study

This article describes the trial design and baseline results for the Silicone Study, a multicenter, randomized surgical trial designed to compare the effectiveness of silicone fluid versus long-acting gas in the treatment of proliferative vitreoretinopathy (PVR). Design features include (1) standardization of the surgical protocol to reduce intersurgeon variability, (2) formulation of a PVR clinical classification system relevant to modern vitreoretinal surgery, and (3) creation of a photographic protocol to document PVR pathology. Statistical issues affecting the analysis of the outcome data include (1) the addition of a second group of patients with more severely diseased eyes after the trial began, (2) the change to a different long-acting gas during the course of the trial, and (3) recurrent retinal detachments that require reoperations with the randomized treatment, and, in some instances, a crossover from the randomized to the alternate treatment. Demographic and baseline ocular characteristics are presented for the two groups under study.

Pars plana vitrectomy for acute retinal detachment in penetrating ocular injuries

We studied 41 eyes with acute retinal detachment after penetrating ocular trauma in a retrospective cohort analysis. Pars plana vitrectomy was performed in 28 eyes, while the remaining 13 eyes had only primary repair and closure of the wound. The two groups differed in the type of trauma (more gunshot wounds in the vitrectomy group and more blunt injuries in the nonvitrectomized group). Visual success (visual acuity of 5/200 or better) was observed in 10 (37%) of the eyes treated by vitrectomy compared with one (8%) of the eyes in the nonvitrectomy group. Anatomic success was achieved in 21 (75%) of the eyes in the vitrectomy group but in only one (8%) of those in the nonvitrectomy group. Enucleation or phthisis was observed in seven (54%) of the eyes in the nonvitrectomy group compared with only five (18%) in the vitrectomy group. Significant prognostic factors for anatomic outcome in the vitrectomy group were the location of the laceration and the presence of the lens.

The effect of therapeutic treatment crossovers on the power of clinical trials. The Silicone Study Group

The Silicone Study is a randomized clinical trial comparing two surgical methods for reattaching the retina when detachment is associated with proliferative vitreoretinopathy. If the retina redetaches subsequently, the patient will usually undergo additional surgery using the assigned treatment. In a limited number of cases the patient may be switched to the alternative treatment, if a "therapeutic crossover" is endorsed by an independent committee of ophthalmologists. A successful outcome is continued anatomic attachment of the retina and an adequate visual result 6 months after the final surgery. The therapeutic treatment crossovers affect the power of the trial to detect a difference between the two treatments. A simulation study shows that the loss in power depends on the magnitude and degree of bias in the probability of being switched from one treatment to the other. Unlike the usual case of lack of compliance, complete information about treatment history may allow statistical adjustment for the crossover. The outcome may be modeled using a multinomial distribution. Much of the power lost due to switching may be recouped under some strong assumptions.

Ocular trauma in an urban population. Review of 1132 cases

A retrospective survey was done of all ocular and adnexal trauma cases seen at a large metropolitan hospital during a 6-month period. By determining patient demographics, causes of the eye injuries, and extent of ocular damage, the authors hoped to delineate areas where preventive measures might decrease such trauma. Demographic and clinical data on 1132 patients were analyzed. Most patients were in the first three decades of life and were male. Blunt trauma was the most common type of injury. Assault was the most common cause and accounted for the highest number of serious injuries. Eye injuries associated with violence are difficult to prevent using conventional strategies that are usually effective in the home and workplace.