Synthesis Of Knowledge Of The Incubation Period Distribution

The main complexities in the analysis and interpretation of epidemiological studies of the incubation period include uncertainty in the dates of infection and the sampling criteria by which individuals are included in the data set. Synthesizing and comparing estimates across studies is important since the estimates are typically based on various methodologies with different underlying assumptions.

Reviews of studies of the incubation period distribution have been given by Moss and Bacchetti (1989) and by Gail and Rosenberg (1992). Table 4.2 presents several estimates of the incubation period distribution reported from a number of different studies that were selected to show a range of methodologies and populations. These studies include: (a) midpoint imputation for the San Francisco City Clinic Cohort of homosexual men; (b) deconvolution of AIDS incidence data in San Francisco; (c) parametric modeling of a hemophilia cohort; and (d) midpoint imputation for a cohort of intravenous drug users in Italy. Several of these studies have identified a significant age effect. For example Goedert, Kessler, Aledort, et al. (1989) found that hemophiliacs over the age of 30 at infection are at higher risk of progression to AIDS than individuals 19 to 30 years old at the time of infection. Similar results were found among hemophiliacs by Darby, Rizza, Doll, et al. (1989) and in a cohort of intravenous drug users (Mariotto, Mariotti, Pezzotti, et al. 1992; see also Italian Seroconversion Study, 1992). Cofactors such as age are considered again in Chapter 5.

A general picture emerges from Table 4.2. First, the probability of developing AIDS within the first two years of seroconversion is very small, less than .03. Then, the hazard of progression to AIDS begins to rise rapidly so that the cumulative probability of developing AIDS within 7.0 years of seroconversion is approximately .25. The cumulative probability of AIDS approaches .50 at 10.0 years following seroconversion.

A striking feature of Table 4.2 is the similarity of the estimates in columns (a)-(d), especially since the modes of viral transmission are

Table 4.2 Cumulative Probabilities of Progressing to AIDS, F[t)

Years Following Seroconversion (i)

SFCC" (midpoint)

San Francisco' (deconvolution)

Hemophiliacs1 (joint parametric modeling)

IVDU^ (midpoint)

1.0

.002

.002

.002

2.0

.010

.009

.012

.015

3.0

.031

.033

.031

4.0

.040

.074

.066

.048

5.0

.135

.113

.093

6.0

.20

.208

.174

.143

7.0

.290

.245

.212

8.0

.37

.371

.325

9.0

.445

.411

10.0

.51

.512

.498

11.0

.54

"San Francisco City Clinic Cohort of homosexual men enrolled in hepatitis vaccine study. Date of seroconversion estimated by midpoint imputation (Hessol, Lifson, O'Malley, et al., 1989).

''Dr convolution of AIDS incidence data in San Francisco using epidemiologic surveys to reconstruct the distribution of infection times (Bacchetti and Moss, 1989). cJoint parametric modeling of 458 hemophiliacs with Weibull model for F(t). F(t) = 1 — exp(—.0021/2'316) for hemophiliacs >age 20 (Brookmeyer and Goedert, 1989).

'Cohort of 468 seroconverters who were injecting drug users (Italian Seroconversion Study, 1992). Additional details of this study in Rezza, Lazzarin, Angarano, et al. (1989).

different. Mariotto, Mariotti, Pezzotti, et al. (1992) compared the incubation distribution among male homosexuals and intravenous drug users in Italian cities and did not find a significant difference.

Follow-up was restricted to about 10 years after infection in these studies. Thus estimates of the incubation period distribution beyond 10 years, and estimates of the mean incubation period, depend on assumptions about how to extrapolate the incubation distribution or upon assumptions about the validity of parametric models beyond the range of the data. Only additional follow-up on these cohorts will further define the tail of the distribution. However, the use of treatments such as zidovudine in these cohorts may alter the incubation period distribution (see Section 8.6, and Chapter 11). Zidovudine was introduced in 1987 for both AIDS patients and infected individuals without AIDS but with advanced HIV disease. Treatments such as zidovudine, inhaled pentamidine and other therapeutic advances may lengthen the incubation period. Accordingly, there will be little

Figure 4.8 Smoothed estimates of the hazard of AIDS using follow-up through January 1, 1987. (Source: Gail and Rosenberg, 1992.) Key:

6Biggar and the International Registry of Seroconverters (1990), excluding those in curve (a) cBacchetti (1990)

Figure 4.8 Smoothed estimates of the hazard of AIDS using follow-up through January 1, 1987. (Source: Gail and Rosenberg, 1992.) Key:

6Biggar and the International Registry of Seroconverters (1990), excluding those in curve (a) cBacchetti (1990)

available information on the natural history of HIV infection (i.e., no treatment intervention) beyond 10 years.

Gail and Rosenberg (1992) evaluated and synthesized information on the hazard of AIDS following seroconversion from various studies. They attempted to eliminate the possible effects of treatment by restricting follow-up to calendar time before 1987. They obtained smooth hazard estimates by fitting spline models (Figure 4.8) to data from: (1) the San Francisco City Clinic Cohort of homosexual men who participated in hepatitis B vaccine trials (Hessol, Lifson, O'Malley, et al. (1989); (2) a cohort of hemophiliacs who attended hemophilia treatment centers (Goedert, Kessler, Aledort, et al., 1989); and (3) an international registry of seroconverters (Biggar and the International Registry of Seroconverters, 1990). In addition, they included hazard estimates by deconvolving AIDS incidence and infection rate estimates in San Francisco (Bacchetti, 1990), as described in Section 4.6. Each of those studies indicates a rapidly increasing hazard up to about year 5 (Figure 4.8). There is little information about the hazard of progression to AIDS, in the absence of treatment, beyond 6 years.

0 0

Post a comment