The UK Register of HIV Seroconverters:
estimating the times from HIV seroconversion to the development of AIDS and death and associated factors
from a cohort of HIV seroconverters
This work is presented as a thesis for the degree of
DOCTOR OF PHILOSOPHY
in
Epidemiology
at the
Faculty of Clinical Sciences
by
Kholoud Porter
From
the Medical Research Council HIV Clinical Trials Centre
University College London Medical School
The Mortimer Market Centre
March 1998 ProQuest Number: U642762
All rights reserved
INFORMATION TO ALL USERS The quality of this reproduction is dependent upon the quality of the copy submitted.
In the unlikely event that the author did not send a complete manuscript and there are missing pages, these will be noted. Also, if material had to be removed, a note will indicate the deletion. uest.
ProQuest U642762
Published by ProQuest LLC(2015). Copyright of the Dissertation is held by the Author.
All rights reserved. This work is protected against unauthorized copying under Title 17, United States Code. Microform Edition © ProQuest LLC.
ProQuest LLC 789 East Eisenhower Parkway P.O. Box 1346 Ann Arbor, Ml 48106-1346 ABSTRACT Knowledge of the distribution of intervals from HIV infection to the development of AIDS and to death, and the factors affecting these intervals is vital to an understanding of the natural history of HIV infection and for making projections of future numbers of AIDS cases. This distribution may have changed since the beginning of the epidemic due particularly to the introduction of anti-retroviral treatment and prophylaxis for Pneumocystis carinii pneumonia. It is likely to be influenced by new advances in the management of HIV infected individuals in the future. Changes in the incubation period distribution could also occur in the absence of changes in available treatments and treatment uptake, due to the evolving distributions of new viral strains. It is therefore important to monitor whether there are changes in the incubation period distribution and, if so, the extent of those changes and factors associated with them. A number of studies have provided estimates for this period in different population groups. Most have tended to focus on one transmission category, e.g. homosexual men, injecting drug users, or haemophiliacs; are small in size; or are no longer recruiting new subjects. This thesis reports on the design, methods and findings from a register of HIV-infected individuals in the UK in whom the date of seroconversion is known with reasonable precision (seroconverters). Baseline and annual follow up information is collected and includes: sex, age, ethnic group, route of HIV transmission, latest CD4 count, details of therapy and prophylaxis, AIDS defining events and vital status. Findings presented in this work are on 2022 seroconverters reported by the end of September 1997, the first 3 years of the Register. Careful documentation of the time that each seroconverter came under unbiased follow up for the purposes of the Register was made so as to minimise any bias resulting from the preferential inclusion of long-term non-progressors and recent seroconverters. Censoring at the end of June 1995, the estimated median time to AIDS from HIV seroconversion was 9.26 years (95% CI= 8.46- 10.39). Censoring at the end of December 1996 the estimated median time from seroconversion to death (from any cause) was 10.79 years (95% CI= 9.81- 11.36 years). Age was found to be highly associated with disease progression with a relative risk of 1.43 (95% CI= 1.19- 1.71) and 1.51 (95% CI= 1.29- 1.78) for each 10-year increase in age to AIDS and death respectively. After adjustment for the effects of other covariates, no evidence of a difference was found between exposure categories, sex or over calendar time. ACKNOWLEDGEMENTS
This work was undertaken while I was employed at the MRC HIV Clinical Trials Centre as Project Co-ordinator for the UK Register of HIV Seroconverters. It was never as a one-woman-band and I am deeply indebted to the many friends and colleagues without whose help, support, encouragement and example I could not have completed this work.
I would especially like to thank Professor Anne Johnson and Professor Andrew Phillips whom I have been fortunate to have as my supervisors. I am extremely grateful to them for always finding time through their busy schedules to give guidance, encouragement and support, and always with patience and good humour.
I am extremely grateful to Professor Janet Darbyshire for her endless support and encouragement throughout my work and study. I am especially thankful to her for being such an approachable and thoughtful person.
I am most indebted to Dr Noël Gill for his guidance over many years and for his faith in my abilities. I am particularly indebted to him for arranging CDSC’s funding of this doctorate.
So many colleagues at CDSC, SCIEH and the Trials Centre have helped make my job less stressful and more manageable and the goals more achievable. I should like to express my gratitude to them all and in particular: CDSC: Dr Barry Evans, Miss Amanda Wright, Mr Dominic Howitt, Dr Ruth Gilbert, Ms Anna Molesworth, and Mrs Pauline Kaye. SCIEH: Dr David Goldberg, Mr Glen Codere, and Ms Geraldine Bums. Clinical Trials Centre: Ms Charlotte Duff, Ms Joanne Gillett, Mr Patrick Kelleher, Ms Sarah Walker, Dr Abdel Babiker, Dr Malcolm Hooker, Mr Adrian Kimberley, and Mr Robert Manning.
This work would not have been possible without the help of clinicians, nurses, database administrators, microbiologists and virologists in a number of centres throughout the United Kingdom. I am very grateful to each of them for their continued help and support. Their names appear in Appendix II to this work.
3 Last, and certainly not least, a special thank you to Tariq and Adam for being the constant source of moral support. CONTENTS
Page
List of tables 6 List of figures 9 Chapter I Introduction 11 Chapter II A review of publications of HIV seroconverter cohorts 30 Chapter III Methods 68 Chapter IV Methodological issues 81 Chapter V Characteristics of the UK Register cohort 98 Chapter VI Statistical methods and an evaluation of bias 118 Chapter VII Progression estimates from HIV seroconversion to 131 AIDS and to death Chapter VIII A summary of findings and further work 153 APPENDICES 165 I Members of the Executive and Steering committees II Names of collaborators III Definition of an acute seroconverter IV Data collection forms V Time of entry into risk set for each clinical centre VI Epidemiology & Infection publication REFERENCES 189 LIST OF TABLES Table Title Page
1.1 Summary of the 1986 classification system for the clinical 15 manifestations of HIV infection 1.2 Summary of the 1993 classification system for the clinical 16 manifestations of HIV infection 1.3 Diseases included in the current definition for AIDS 18 1.4 Definitive Diagnostic Methods for Diseases Indicative of AIDS 19 1.5 Presumptive Methods for Diseases Indicative of AIDS 20 1.6 AIDS cases and HIV infection reports in the UK: data reported to 22 CDSC and SCIEH to the end of September 1994 1.7 Drugs prescribed in HIV infection 27 2.1 A summary of published studies of HIV progression rates in cohorts of 33 homosexual and bisexual men 2.2 A summary of published studies of HIV progression rates in cohorts of 44 persons with haemophilia 2.3 Estimates of progression to AIDS by age group at seroconversion from 46 the UKHCDO data 2.4 Estimates of progression to AIDS within 9 years of seroconversion by 46 age group for persons in the MCHS 2.5 A summary of published studies of blood transfusion recipients 49 2.6 Progression to AIDS by age group at seroconversion for Swedish 53 transfusion recipients 2.7 A summary of published studies of injecting drug users 55 2.8 A comparison of progression rates to AIDS between haemophiliacs in 61 Western Pennsylvania and homosexual men in the SFCC 2.9 A comparison of progression rates to AIDS between haemophiliacs in 61 MCHS and homosexual men in the 1RS 2.10 Progression rates to AIDS in haemophiliacs and transfusion recipients 64 in Sweden 4.1 Exposure category distribution of seroconverters, reported AIDS cases 83 and HIV infections (to the end of September 1994) and estimated prevalence (at the end of 1993) 4.2 Late entry: number of patients recruited from each of 3 hypothetical 90 clinics 4.3 Late entry: life table of retrospectively identified seroconverters from 90 hypothetical clinics A, B, and C (see Table 4.2) 4.4 Reporting delay for UK AIDS cases reported to CDSC and SCIEH to the 95 end of 1996 5.1 Characteristics of all subjects reported to the Register 100 October 1994 - September 1997 5.2 Total number of all subjects reported, number of AIDS diagnoses, and 104 number of deaths known by 30 September 1997 5.3 Age distribution of all subjects reported to the Register 105 by exposure category 5.4 Age at seroconversion of all subjects reported to the Register by 106 exposure category and sex 5.5 Interval between last negative and first positive HIV antibody tests and 109 proportions with AIDS and who have died 5.6 AIDS cases: Year of diagnosis of AIDS 110 5.7 AIDS cases: AIDS defining diseases and proportion of total diagnoses 112 5.8 AIDS cases: exposure category distribution of all AIDS-defining 113 conditions 5.9 Non-progressors: year of last clinic visit for persons not diagnosed 114 with AIDS 5.10 Deaths : Estimated year of seroconversion by year of death 115 5.11 Persons remaining alive in HIV survival estimates: year last known to 117 be alive 6.1 Documentation of seroconversion by exposure category for 122 seroconverters with a 3-year maximum HIV test interval and proportion of total reported 6.2 Characteristics of 917 documented seroconverters with 3 year maximum 124 test interval and contributing information to the risk set of AIDS analysis 6.3 Characteristics of 1149 documented seroconverters with 3 year 125 maximum test interval and contributing information to the risk set of HIV survival analysis 6.4 Examination of the effect of ignoring late entry on AIDS progression 126 estimates 7.1 A descriptive comparison between: N- the total numbers of subjects 133 reported to the Register (n= 2022) and the numbers included in the AIDS (n= 917) (A) and death (n= 1149) (D) progression analyses 7.2 Cumulative proportion of seroconverters developing AIDS 135 7.3 F actors associated with progression from HIV seroconversion to an 137 AIDS diagnosis 7.4 Examination of the effect of age at seroconversion on progression 139 estimates from HIV seroconversion to AIDS 7.5 Examining the effect of age at seroconversion on AIDS diagnosis for 141 men exposed through sex between men 7.6 The effect of narrower HTV test intervals on AIDS progression estimates 144 within the first 3 years of HIV seroconversion 7.7 Cumulative proportion of seroconverters dying from any cause following 147 HIV seroconversion 7.8 Factors associated with progression from HIV seroconversion to death 149 7.9 Examining the effect of age at seroconversion on survival estimates 150 7.10 The effect of narrower HIV test intervals on HIV survival estimates 152 within the first 3 years of HIV seroconversion 8.1 The relative risk of AIDS for injecting drug users compared to men who 156 have sex with men: findings from published studies 8.2 The relative risk of AIDS of a 10-year increase in age at seroconversion 158 from published studies LIST OF FIGURES Figure Title Page
1.1 WHO estimates of the number of persons living with HIV at the end 23 of 1996, and estimates of the cumulative number of AIDS cases in adults to the end of 1996 2.1 Published progression rates from HIV seroconversion to AIDS from 34 cohorts of homosexual men 2.2 Published progression rates to AIDS from cohorts of injecting drug 56 users 4.1 An illustration of right censoring bias 92 5.1 Distribution of all subjects reported to the Register 99 to the end of September 1997 by reporting clinical centre 5.2 Exposure category distribution by calendar year of seroconversion 102 5.3 Age distribution of all subjects reported by estimated year of 107 seroconversion 6.1 Kaplan-Meier plots of time to AIDS from seroconversion for 122 documented and undocumented seroconverters 6.2 Kaplan-Meier plots of time to AIDS from seroconversion allowing for 126 late entry and ignoring it 6.3 Kaplan-Meier plots of the effect of right-censoring strategies on AIDS 129 progression estimates 6.4 Kaplan-Meier plots of the effect of right-censoring strategies on HIV 130 survival estimates 7.1 A flow chart of number of subj ects reported to the Register and 132 numbers included in final analyses 7.2 Cumulative progression rates from HIV seroconversion to AIDS 136 diagnosis for seroconverters reported to the Register and censored in June 1995 with persons lost to follow up assumed to be AIDS-free on that date 7.3 Kaplan-Meier plot for time to AIDS from HIV seroconversion by age 139 group at HIV seroconversion 7.4 Kaplan-Meier plots for time to AIDS from HIV seroconversion by age 141 group at HIV seroconversion for men exposed through sex between men 7.5 Estimated AIDS rate and 95% confidence intervals for each year since 143 seroconversion 7.6 Kaplan-Meier plots of the effect of the HIV test interval on progression 144 estimates from HIV seroconversion to AIDS 7.7 Cumulative progression rates from HIV seroconversion to death for 146 seroconverters reported to the Register and censored in December 1996 with persons lost to follow up assumed to be alive on that date 7.8 Kaplan-Meier plots for time from seroconversion to death from any 150 cause by age group at seroconversion 7.9 Kaplan-Meier plots of the effect of the HIV test interval on progression 152 estimates from HIV seroconversion to death 8.1 Cumulative proportion of persons developing AIDS by time since 154 seroconversion: estimates from the Register compared to estimates from other studies of HIV seroconverters
10 CHAPTER I
INTRODUCTION
1.1 First AIDS reports
An article appeared in the Morbidity and Mortality Weekly Report (MMWR) from the Centers for Disease Control (CDC) in Atlanta, USA in June 1981 describing a cluster of 5 cases of Pneumocystis carinii pneumonia (PCP) in previously fit, healthy young homosexual men in Los Angeles without any apparent reason for immune suppression ’. One month later 10 more cases of PCP and 26 of Kaposi’s sarcoma (KS) from New York as well as California were described
PCP, a lung infection caused by a pathogen to which most individuals are exposed with no undue consequences, was extremely rare prior to 1981 in individuals other than those receiving immunosuppressive therapy or among the ehronically malnourished. In 1967, CDC became the sole supplier in the United States of pentamidine isethionate, then the only recommended PCP therapy, and began collecting data on each PCP case diagnosed and treated in the US. After reviewing requests for pentamidine in the period 1967 to 1970, researchers found only one case of confirmed PCP without a known underlying condition In the period immediately prior to the recognition of AIDS, January 1976 to June 1980, CDC received only one request for pentamidine isethionate to treat an adult in the United States who had PCP and no underlying cause of immune suppression. In 1981 alone, 42 requests for pentamidine were received to treat patients with PCP and no known underlying disorders
KS, a very rare skin neoplasm, had until the report in the MMWR in 1981 affected mostly older men of Mediterranean origin or cancer or transplant patients undergoing immunosuppressive therapy
The widespread occurrence of KS and PCP in young people with no underlying disease
11 or history of immunosuppressive therapy was unprecedented. Searches of the medical literature, autopsy records and tumour registries revealed that these diseases previously had occurred at very low levels in the United States
It soon became evident that these young homosexual men had a common immunologic deficit, an impairment in cell-mediated immunity resulting from a significant loss of "T- helper" cells, which bear the CD4 marker.
Soon after the initial reports from the United States other countries reported cases of what later became known as AIDS (Acquired Immune Deficiency Syndrome). The fact that homosexual men constituted the initial population in which AIDS occurred in the United States led epidemiologists to conclude that a homosexual lifestyle was specifically related to the disease These early suggestions were dismissed when the syndrome was later observed in distinctly different groups: in male and female injecting drug users in haemophiliacs and blood transfusion recipients among female sex partners of bisexual men recipients of blood or blood products or injecting drug users and among infants bom to mothers with AIDS or with a history of injecting dmg use 20-22
The first case in the United Kingdom was described in 1981 and in September 1982 a surveillance system was set up, through a voluntary reporting scheme, of Kaposi’s sarcoma and AIDS jointly by the Public Health Laboratory Service (PHLS) Communicable Disease Surveillance Centre (CDSC) and the Communicable Disease (Scotland) Unit (CD(S)U) (now known as The Scottish Centre for Infection and Environmental Health (SCIEH))
1.2 Isolation of responsible agent
Many public health experts concluded that the clustering of AIDS cases and the oceurrence of cases in diverse risk groups could be explained only if AIDS were caused by an infectious micro-organism transmitted in the manner of hepatitis B vims (HBV): by sexual contact, by inoculation with blood or blood products, and from mother to new-born infant.
12 By 1983, several research groups had focused on retroviruses for clues to the cause of AIDS. Two recently recognised retroviruses, human T-cell lymphotropic virus (HTLV) type-I and HTLV-II, were the only viruses then known to preferentially infect helper T lymphocytes, the cells depleted in people with AIDS. The pattern of HTLV transmission was similar to that seen among AIDS patients. In addition, HTLV-I was known to cause mild immunosuppression, and a related retrovirus, the lymphotropic feline leukaemia virus (FeLV), caused lethal immunosuppression in cats.
In May 1983, the first report providing experimental evidence for an association between a retrovirus and AIDS was published After finding antibodies cross-reactive with HTLV-I in a homosexual patient with lymphadenopathy, a group led by Dr. Luc Montagnier in France isolated a previously unrecognised virus producing reverse transcriptase which later became known as lymphadenopathy-associated virus (LAV). The French group subsequently reported that LAV was tropic for T-helper cells, in which it grew to substantial titres and caused cell death.
In 1984, a considerable amount of new data added to the evidence for a retroviral aetiology for AIDS. Researchers at the National Institutes of Health reported the isolation of a cytopathic T-lymphotropic virus from 48 different people, including 18 of 21 with pre-AIDS, three of four clinically normal mothers of children with AIDS, 26 of 72 children and adults with AIDS, and one (who later developed AIDS) of 22 healthy homosexuals The virus, named HTLV-III, could not be found in 115 healthy heterosexual subjects. Later, in 1986, the International Committee of Viral Taxonomy renamed these identical viruses the human immunodeficiency virus (HIV)
1.3 Characterisation of HIV disease
HIV selectively infects and destroys an integral part of the human immune system - the white blood cells, in particular the T4+ lymphocytes (or CD4 cells) because of the affinity of the virus for the CD4 surface marker These so-called T-helper cells co ordinate a number of critical immunologic functions- their loss results in the progressive impairment of the immune system and is associated with a deteriorating clinical course. Infected cells (or the new virus particles they produce), can then be transmitted to those
13 uninfected. In advaneed HIV disease, abnormalities of virtually every component of the immune system are evident. As disease progresses, inereasing amounts of infectious virus, viral antigens and HIV-specific nucleic acids in the body correlate with a worsening clinical course. HIV disease progression can be described in terms of stages defined by clinical manifestations, as proposed by CDC in 1986 (Table 1.1), and later superseded by a system incorporating the infected individual’s CD4 cell count (Table 1.2).
Initial infection may or may not be accompanied by an acute illness (called seroconversion illness) which resembles glandular fever A persistent swelling of the lymph glands may also be diagnosed (PGL). A period of asymptomatic infection of variable length follows which is accompanied by a general decline of the immune system. The final stage is the diagnosis of AIDS. This is a spectrum of conditions that occur when the immune system is seriously damaged leading to severe immune suppression and constitutional disease, neurological complications, and opportunistic infections and neoplasms that rarely occur in persons with intact immune function.
1.4 AIDS case definition changes
The first case definition, set up in 1982, was in reality for ‘Kaposi’s sarcoma and opportunistic infections in previously healthy persons’:
“A case is defined as illness in a person who 1) has either biopsy-proven KS or culture-proven, life threatening opportunistic infection, 2) is under age 60 and 3) has no history of either immunosuppressive underlying illness or immunosuppressive therapy ”
Three months later when the syndrome was termed AIDS, a case definition, which became widely adopted for the purposes of surveillance, was defined as someone with:
“a disease at least moderately predictive of a defect in cell-mediated immunity, occurring in a person with no known cause for diminished resistance to that disease. Such diseases include KS, PCP and serious other opportunistic infections. Diagnoses are considered to fit the case definition only if based on sufficiently reliable methods (generally histology or culture). ”
14 Table 1.1 Summary of the 1986 classification system for the clinical manifestations of HIV infection
Classification Disease/ symptoms Group I Acute infection Group II Asymptomatic^ Group III Persistent generalised lymphadenopathy (PGL)^ Group IV AIDS and ARC Subgroup A Constitutional disease Subgroup B Neurological disease Subgroup C Secondary infectious disease Category C-1 Specified secondary infectious diseases: PCP, chronic cryptosporidiosis, toxoplasmosis, extraintestinal strongyloidiasis, isosporiasis, candidiasis (oesophageal, bronchial, or pulmonary), cryptococcosis, histoplasmosis, mycobacterial infection with Mycobacterium avium complex or M, kansasii, cytomegalovirus infection, chronic mucocutaneous or disseminated herpes simplex virus infection, or progressive multifocal leukoencephalopathy. Category C-2 Other specified secondary infectious diseases: oral hairy leukoplakia, multidermatomal herpes zoster, recurrent salmonella bacteraemia, nocardiosis, tuberculosis, or oral candidiasis. Subgroup D Secondary cancers: Kaposi's sarcoma, non-Hodgkin's lymphoma (small, noncleaved lymphoma or immunoblastic sarcoma), or primary lymphoma of the brain. Subgroup E Other conditions
^ Patients in Groups II and III may be subclassified on the basis of a laboratory evaluation.
15 Table 1.2 Summary of the 1993 classification system for the clinical manifestations of HIV infection
Clinical categories (A) (B) (C) CD4+ T-cell Asymptomatic, acute Symptomatic, not (A) AIDS-indicator categories (primary) HIV or PGLf or (C) conditions§ conditions^ (1) >500/^1 A1 B1 Cl
(2) 200-499/|Lil A2 B2 C2 (3) <200/|nl A3 B3 C3 AIDS-indicator T-cell count
t PGL= persistent generalised lymphadenopathy. Clinical category A includes acute (primary) HIV infection with accompanying illness or history of acute HIV infection Conditions listed in Categories B and C must not have occurred. § Includes symptomatic HIV infected adults with conditions not included in clinical Category C and who meet at least one of the following criteria: a) the conditions are attributed to HIV infection or are indicative of a defect in cell-mediated immunity; or b) the conditions are considered by physicians to have a clinical course or to require management, for example, bacillary anginomatosis, oropharyngeal candidiasis, cervical dysplasia, constitutional symptoms such as fever or diarrhoea lasting more than one month, oral hairy leukoplakia. For classification purposes. Category B conditions take precedence over those in Category A. ^ Includes the clinical conditions listed in the surveillance case definition (see Table 1.3). For classification purposes, once a Category C condition has occurred, the person will remain in Category C.
16 After the discovery of HIV, a revision of the case definition was proposed by CDC in 1985 which was adopted in industrialised countries adding the diagnosis of non- Hodgkin’s lymphoma and also the diagnosis of KS in a person over the age of 60 with a positive serologic or virologie test for HIV- previously a KS diagnosis was restricted to persons aged less than 60 years
In 1987 a major revision followed which increased both sensitivity and specificity of the case definition by allowing the presumptive non-invasive diagnosis of many AIDS indicator diseases together with laboratory evidence of HIV infection Two new diseases were also incorporated into the definition: HIV Wasting Syndrome and HIV Encephalopathy (Tables 1.3 - 1.5).
The most recent revision was developed in 1993 and widened the 1987 definition to include 3 new diseases: pulmonary tuberculosis (TB), recurrent pneumonia in a 12- month period, and invasive cervical carcinoma, as well as a CD4 cell count of 200 or less per pi of blood regardless of clinical manifestations This revision was adopted in the UK but excluded the diagnosis of severe immune suppression (i.e. a CD4 cell count of below 200 cells per pi of blood) in the absence of clinical illness
Surveillance definitions of AIDS have proven useful epidemiologically to track and quantify the epidemic of HIV-mediated immunosuppression and its manifestations. However, AIDS represents only the end stage of a continuous, progressive pathogenic process. In clinical practice, symptoms together with the measurements of immune function, particularly levels of CD4+ T lymphocytes, are used to guide the treatment of HIV-infected persons.
In many developing countries, where diagnostic facilities may be minimal, epidemiologists employ a case definition based on the presence of various clinical symptoms associated with immune deficiency and the exclusion of other known causes of immunosuppression, such as cancer or malnutrition.
17 Table 1.3 Diseases included in the current definition for AIDS
Candidiasis of bronchi, trachea, or lungs Candidiasis, oesophageal Cervical cancer, invasive** Coccidioidomycosis, disseminated or extrapulmonary Cryptococcosis Cryptosporidiosis, chronic intestinal (>1 month duration) Cytomegalovirus disease (other than liver, spleen or nodes) Cytomegalovirus retinitis (with loss of vision) Encephalopathy, HIV related* Herpes simplex: chronic ulcer(s) (>1 month duration); or bronchitis, pneumonitis or oesophagitis Histoplasmosis, disseminated or extrapulmonary Isosporiasis, chronic intestinal (> 1 month duration) Kaposi’s sarcoma Lymphoma, Burkitt’s or equivalent* Lymphoma, immunoblastic or equivalent* Lymphoma, primary of the brain* Mycobacterium avium complex or Mkansaii, disseminated or extrapulmonary Mycobacterium tuberculosis, any site (pulmonary** or extrapulmonary*) Mycobacterium, other species or unidentified species, disseminated or extrapulmonary Pneumocystis carinii pneumonia Pneumonia, recurrent** Progressive multifocal leukoencephalopathy Salmonella septicaemia, recurrent Toxoplasmosis of brain Wasting syndrome due to HIV*
* added in 1987 ** added in 1993
18 Table 1.4 Definitive Diagnostic Methods for Diseases Indicative of AIDS The following list indicates the definitive diagnostic methods for diseases indicative of AIDS. These must be accompanied with laboratory evidence of HIV infection regardless of the presence of other causes of immunodeficiency:
Cryptosporidiosisf Microscopy (histology or cytology) Isosporiasis KSff, Lymphoma of brainfî, other non-Hodgkin’s lymphoma PCPf, PMLt Toxoplasmosis of braint Cervical cancer Candidiasisf Gross inspection by endoscopy or autopsy or by microscopy (histology or cytology) on a specimen obtained directly from the tissues affected (including scrapings from the mucosal surface), not from a culture. Coccidioidomycosis Microscopy (histology or cytology), culture, or detection of antigen in a Cryptococcosisf specimen obtained directly from the tissues affected or a fluid from those Cytomegalovirus^ tissues. Herpes simplex virusf Histoplasmosis Tuberculosis Culture Other mycobacteriosis Salmonellosis HIV encephalopathy Clinical findings of disabling cognitive or motor dysfunction interfering with occupation or activities of daily living, progressing over weeks to months, in the absence of a concurrent illness or condition other than HIV infection that could explain the findings. Methods to rule out such concurrent illness and conditions must include cerebrospinal fluid examination and either brain imaging (computed tomography or magnetic resonance) or autopsy. HIV wasting syndrome Findings of profound involuntary weight loss of greater than 10 percent of baseline body weight plus either chronic diarrhoea (at least two loose stools per day greater than or equal to 30 days), or chronic weakness and documented fever (for greater than or equal to 30 days, intermittent or constant) in the absence of a concurrent illness or condition other than HIV infection that could explain the findings (e.g., cancer, tuberculosis, cryptosporidiosis, or other specific enteritis). Pneumonia, recurrent Recurrent (more than one episode in a 1-year period), acute (new x-ray evidence not present earlier) pneumonia diagnosed by both (a) culture (or other organism-specific diagnostic method) obtained from a clinically reliable specimen of a pathogen that typically causes pneumonia (other than Pneumocystis carinii or Mycobacterium tuberculosis), and (b) radiologic evidence of pneumonia; cases that do not have laboratory confirmation of a causative organism for one of the episodes of pneumonia will be considered to be presumptively diagnosed. t These diseases are also included in the absence of laboratory evidence of HIV infection but after the exclusion of other causes of immunodeficiency.
Î Without laboratory evidence of HIV infection the inclusion of KS and lymphoma of the brain is limited to persons below 60 years of age.
19 Table 1.5 Presumptive Methods for Diseases Indicative of AIDS The following list are suggested criteria for a presumptive diagnosis of diseases indicative of AIDS, regardless of the presence of other causes of immunodeficiency, (these must be accompanied by laboratory evidence of HIV infection);
Candidiasis of Recent onset of retrosternal pain on swallowing; AND (b) oral Oesophagus candidiasis diagnosed by the gross appearance of white patches or plaques on an Erythematous base or by the microscopic appearance of fungal mycelial filaments fi-om a noncultured specimen scraped from the oral mucosa Cytomegalovirus characteristic appearance on serial ophthalmoscopic examinations (e.g., Retinitis discrete patches of retinal whitening with distinct borders, spreading in a centrifugal manner along the paths of blood vessels, progressing over several months, and frequently associated with retinal vasculitis, haemorrhage, and necrosis). Resolution of active disease leaves retinal scarring and atrophy with retinal pigment epithelial mottling Mycobacteriosis Microscopy of a specimen from stool or normally sterile body fluids or tissue from a site other than lungs, skin, or cervical or hilar lymph nodes that shows acid-fast bacilli of a species not identified by culture KS A characteristic gross appearance of an erythematous or violaceous plaque like lesion on skin or mucous membrane. (Note: Presumptive diagnosis of Kaposi's sarcoma should not be made by clinicians who have seen few cases of it.) PCP A history of dyspnoea on exertion or non-productive cough of recent onset (within the past 3 months); AND (b) chest x-ray evidence of diffuse bilateral interstitial infiltrates or evidence by gallium scan of diffuse bilateral pulmonary disease; AND (c) arterial blood gas analysis showing an arterial pO of < 70 mm Hg or a low respiratory diffusing capacity (< 80% of predicted values) or an increase in the alveolar- arterial oxygen tension gradient; AND (d) no evidence of bacterial pneumonia. Toxoplasmosis of criteria (a) Recent onset of a focal neurologic abnormality consistent Brain with intracranial disease or a reduced level of consciousness; AND (b) evidence by brain imaging (computed tomography or magnetic resonance) of a lesion having a mass effect or the radiographic appearance of which is enhanced by injection of contrast medium; AND (c) serum antibody to toxoplasmosis or successful response to therapy for toxoplasmosis. Pulmonary When bactériologie confirmation is not available, other reports may be Tuberculosis considered to be verified cases of pulmonary tuberculosis if the criteria of the Division of Tuberculosis Elimination, National Center for Prevention Services, CDC, are used* Pneumonia, Recurrent (more than one episode in a 1-year period) or acute (new recurrent symptoms, signs, or x-ray evidence not present earlier) pneumonia diagnosed on clinical or radiologic grounds by the patient’s physician. available in MMWR 1990; 39 (No. RR-I3):39-40.
20 1.5 Emergence of a disease
By February 1985, 8495 cases of AIDS in the US had been reported to CDC, and in Europe 762 cases were reported to the World Health Organisation (WHO) Collaborating Centre for the European region in Paris. A total of 132 cases were reported from the UK, 126 (95%) of whom were male.
By that time the major modes for transmission of HIV infection had become known: through sexual contact (homosexual and heterosexual), through blood contact (injecting drug use, receipt of contaminated blood products by transfusion or clotting Factor VIII, or occupationally in a health care setting), and from mother to infant (in utero, during birth and through breastfeeding). The major mode of transmission of HIV in the UK is through sex between men.
At the end of September 1994 when this work was embarked upon, 9865 AIDS cases had been reported to CDSC and SCIEH of whom 6712 (68%) were known to have died. As reporting of AIDS cases is not 100% complete due to under-reporting as well as delayed reporting, this figure should be regarded as the minimum cumulative AIDS incidence in the UK. 22,581 laboratory reports of HIV infection were also made to CDSC and SCIEH, of whom a number were also contained in the AIDS figures (Table 1.6). In the absence of regular mandatory testing for HIV antibodies in the population, not all those infected will have been diagnosed. Thus the number of laboratory reports under-estimates the total number of persons infected with HIV in the population. Allowing for under-diagnosis as well as delayed and under-reporting, the estimates for the cumulative number of AIDS cases at the end of 1994 in the UK was 12,170 and 23,880 prevalent HIV infections at the end of 1993
At the end of 1996, WHO estimate the total cumulative number of AIDS cases world wide to be 6.7 million, and of prevalent HIV infections to be 22.6 million (Figure 1.1). Most of these are in sub-Saharan Africa, while the UK is estimated to account for approximately 0.1% of the world total.
21 Table 1.6 AIDS cases and HIV infection reports in the UK: data reported to CDSC and SCIEH to the end of September 1994
Exposure category Male Female Total AIDSHIVAIDSHIV AIDS HIV*
Sex between men 7334 13966 - - 7334 (74%) 13966 (62%) Sex between men & women 675 1689 506 1951 1181(12%) 3646 (16%) Injecting drug use 395 1821 168 820 563 (6%) 2645 (12%) Haemophilia 427 1215 6 11 433 (4%) 1226 (5%) Blood transfusion 37 76 68 77 105 (1%) 154 (0.7%) Other/undetermined 160 687 89 225 249 (3%) 944 (4%)
Total 9028 19454 837 3084 9865 (100%) 22581 (100%)
* Total HIV laboratory reports includes 43 sex not stated
N) K> Figure 1.1 WHO estimates of the number of persons living with HIV at the end of 1996
IT 7$(\opQ) A
^ .y - - 5 .2 i^ i g i i M miHion
WHO estimates of the cumulative number of AIDS cases in adults to the end of 1996
g=À'
53,000.
'million
.700
23 1.6 The incubation period
The incubation period of a disease is defined as the time between the entrance of the pathogen into the host to the development of clinical signs and symptoms.
Shortly after the discovery of HIV and development of an antibody test able to identify persons infected with the virus, estimates of the AIDS incubation period distribution (IPD) were made. It became apparent early in the epidemic that many more persons were infected than those diagnosed and that estimates of the time from seroconversion to disease manifestation as well as information on the factors which may influence this period would be insightful into the pathogenesis of HIV and could lead to better understanding of its natural history. Besides being a valuable tool in the clinical assessment and counselling of those infected, knowledge of the IPD is crucial to healthcare planners as it allows estimates of the future AIDS case load to be made.
However, neither the start of this period nor its end are well-defined events. Once HIV enters a host, the host is said to be infected. However, infection cannot normally be diagnosed until antibodies specific to HIV antigens are detected in the host’s blood. This usually occurs within 6 weeks of infection. This transition from being negative for HIV antibodies to becoming positive is termed ‘seroconversion’ and the host is said to have seroconverted. The AIDS incubation period is generally measured from seroconversion, which is regarded as synonymous with infection. As the time lag between infection and seroconversion is short relative to the incubation period, estimates of the IPD are unlikely to be greatly affected by it.
Early estimates of the IPD were derived from information available on persons with AIDS who had been infected through contaminated blood transfusion. These analyses, however, produced estimates of 4.3 years for the median time to AIDS much shorter than current estimates in industrialised countries. This was almost certainly due to an under-estimate of the size of the unobserved cases of infection, i.e. persons whose HIV infection had not yet been diagnosed, and persons whose incubation period was longer than the period under observation.
24 Adding further complication is the fact that there does not exist one specific definable event in HIV disease which heralds the “development of clinical signs and symptoms”, and many symptoms may remain clinically undiagnosed. Less subject to this bias is the diagnosis of AIDS, and the IPD is thus generally measured to AIDS.
Nonetheless, variation will arise in its diagnosis due to: which case AIDS definition is operable at the time, which AIDS indicator disease is diagnosed (as some tend to present at more severe states of immune suppression than others) and to a lesser extent, the differing experience of diagnosis of the clinician.
A particular characteristic of the AIDS incubation period is its length and variability. The progression from seroconversion to the diagnosis of an AIDS defining disease can be as short as a few months or as long as 15 years or over. We do not yet know how long this tail of the distribution is likely to be nor can we be certain that all those infected will inevitably progress to AIDS. Our knowledge of the AIDS incubation period, current at the time when this work was embarked upon, is that it has a median of the order of 8 to 10 years and is strongly influenced by the age of the infected individual at the time of seroconversion. In adults it now seems clear that younger individuals (below 25- 30 years) will, on the whole, tend to progress at much slower rates than older individuals. Moreover, a number of studies of injecting drug users and persons with haemophilia have produced much slower estimates of the incubation period than studies of homosexual men (discussed in Chapter II). It is not entirely clear how much of this is real and how much may be due to methodological differences. As most cohorts have tended to focus on one exposure group, across-group comparisons between publications may not be appropriate.
In any event, our current understanding of the AIDS incubation period is based largely on data from individuals who seroconverted in the 1980’s (see Chapter II). With the advances in the care and management of infected individuals in the mid-1990’s a number of anti-retroviral drugs is now available to most infected individuals in industrialised countries. This has led not only to hopes of an improvement in survival but also fears over failures with the emergence of drug-resistance.
25 1.7 The UK Register of HIV seroconverters
I began working in October 1994 to set up a register of persons in the UK with well- estimated dates of HIV seroconversion. The Register was initially funded for a period of 3 years through a grant from the Medical Research Council with the aim of providing ongoing monitoring of the AIDS incubation period and of the time from HIV seroconversion to death. Of particular interest were any changes in these periods over calendar time and if these could be due to anti-retroviral treatment and prophylaxis for opportunistic infections. With this in mind, it is worth summarising what treatment was generally available for HIV infected persons.
1.8 Treatment and prophylaxis for HIV and AIDS
Drugs prescribed for the treatment of HIV disease may be used: either against the opportunistic infections which the infected individual becomes susceptible to, or against the virus itself. Table 1.7 summarises the major classes of drugs currently prescribed.
1.8.1 Drugs used against opportunistic infections
These can be prescribed for the prevention (primary prophylaxis), against the recurrence (secondary prophylaxis) or for the treatment of a number of opportunistic infections. In the early 1980s these were the only drugs available for the management of HIV infected individuals.
The most widely used drugs are against PCP, e.g. Pentamidine and Co-trimoxazole, and have had dramatic results in reducing its incidence. Other drugs currently widely prescribed are also summarised in Table 1.7
1.8.2 Drugs used against the virus
These fall into two major categories according to which HIV enzyme the drug targets: Reverse-transcriptase inhibitors (RTIs), and Protease inhibitors (Pis).
26 Table 1.7 Drugs prescribed in HIV infection
Antiretrovirals Drugs against opportunistic infections
RTIs Pis Infection Drug nucleoside non-nucleoside Candida fluconazole analogue analogue itraconazole ketoconazole AZT nevaripine saquinavir CMV disease foscamet ddC delavirdine ritonavir ganciclovir ddl loviride indinavir HSV acyclovir D4T nelfinavir MAI rifabutin 3TC PCP cotrimoxazole fansidar pentamidine IB isoniazid rifampicin to <1 The most widely prescribed of these drugs and the first to be approved for use in HIV infection is Zidovudine (AZT). It is a ‘nucleoside analogue’ RTI, terminating DNA synthesis from viral RNA thus preventing the virus from replicating. ‘Non-nucleoside analogue’ RTIs also target HIV’s reverse transcriptase enzyme by binding onto it to prevent it from working.
The other category of anti-HIV drugs are the protease inhibitors. These target a different enzyme (protease) to generate non-infectious HIV virions. This class of drugs were not generally prescribed in the UK until 1996 or later, and generally as part of a ‘triple regimen’ in combination with other anti-retrovirals, particularly two nucleoside analogue RTIs.
Most of these drugs, however, appear to lack long-term efficacy, which may be partly due to the development of resistance, and are toxic.
1.10 Objectives of this thesis
The aim of this thesis is to measure the incubation period and survival time in a national cohort of individuals with known dates of seroconversion and to explore demographic, clinical and temporal factors which may influence these distributions.
This thesis will:
♦ Describe how the Register was set up and how seroconverters, potentially eligible for inclusion are identified. ♦ Describe how follow up of enrolled subjects is conducted and what methods of minimising loss to follow up are employed. ♦ Describe the characteristics of persons reported to the Register. ♦ Give estimates for cumulative progression rates from HIV seroconversion to AIDS and to death. ♦ Examine the effects of the following cofactors on HIV progression rates:
♦ Exposure category (sex between men, injecting drug use, sex between men and women, other)
28 * Age at estimated calendar time of seroconversion * Calendar year of seroconversion * Sex * HIV test interval (time between last negative and first positive antibody test dates) * How seroconverters were identified
1.11 Layout of thesis
Chapter II reviews the publications summarising knowledge, at the time when the Register was being set up, on the AIDS incubation period. Chapter III details methods: how subjects are identified for inclusion in the Register and how they are followed. Chapter IV discusses methodological issues in estimating the periods from seroconversion to AIDS and to death. The role of bias and its potential sources are explored and how it can affect estimates. Chapter V summarises the cohort in terms of demographic factors, such as age, sex, exposure category, ethnic group, likely country of infection, estimated year of seroconversion and clinical factors including deaths the presence of a possible seroconversion illness, AIDS events and deaths. Chapter VI details the effect of a number of potential sources of bias on data from the Register, and how these have been dealt with in the statistical analyses. Chapter VII examines the AIDS incubation period and survival time, and associated cofactors. Finally, Chapter VIII discusses findings and the contribution made by this thesis to knowledge in the context of other research in this area, and sketches what further work is to be done beyond the scope of this thesis.
29 CHAPTER II
A REVIEW OF PUBLICATIONS OF HIV SEROCONVERTER COHORTS
Much of our present knowledge of the AIDS incubation period is derived from cohorts of persons already infected with HIV when enrolled, who are monitored over time These are termed ‘prevalent cohorts’ and those infected are ‘prevalent positive’. The main advantage of such cohorts to investigators is that they are relatively easy to assemble and contain many events (i.e. the development of AIDS and/or death) over the follow up period. Uncertainty, however, over the period before observation began means that this unobserved period has to be supplemented with information from other sources to impute infection dates. Methods include the use of markers of disease progression or an assumed distribution of infection. Such data are thus ‘left truneated’ and the magnitude of bias in estimates obtained from such cohorts will strongly depend on how soon after infection each person enters the cohort.
In order to reduce this uncertainty over the exact time of infection several investigators have gathered eohorts of initially HIV seronegative persons followed up through seroconversion to AIDS and death. Many of these ‘incident eohorts’ were initially opportunistically assembled from previous case-eontrol studies where they had acted as the ‘controls’ (while those already infected were the ‘eases’) and had seroconverted whilst enrolled. Others were retrospeetively determined through the testing of sera which had been stored for reasons unrelated to HIV infection, such as hepatitis B. These ‘incident infections’ have the advantage of having a more preeisely defined time of infection. Due to the fairly low ineidenee of HIV, those ascertained prospeetively at least, require lengthy periods of follow up. As with prevalent cohorts, retrospectively ascertained ‘incident infections’ may also be left-truncated (discussed in section 4.5).
This chapter summarises the information which was available in Oetober 1994, when the UK Register of HIV Seroconverters was initiated, from publications arising from cohorts of HIV infected individuals. Information summarised here is grouped according to the main exposure eategory of subjects in the cohort: homosexual and bisexual men,
30 persons infected through contaminated clotting factor, Factor VIII, transfusion recipients of infected blood, and injecting drug users. The studies are described in detail below and summary findings are presented in tables and figures for each exposure category. The final section details findings from studies comprising different groups of persons either within the same cohort or through pooling data from more than one cohort.
The studies listed here were identified through searching Medline. The references listed in the publications thus identified were also reviewed. Only English language publications were reviewed. As the main objective of this study is to estimate the time from seroconversion to AIDS and to death I have only included studies from which it was possible to obtain information on these estimates. Studies which only gave estimates from the time of enrolment have not been included.
2.1 COHORTS OF HOMOSEXUAL AND BISEXUAL MEN
Many of the early cohorts were of homosexual and bisexual men as they represented a group of individuals who were either already antibody positive when enrolled (prevalent infections) or, because they were at risk of HIV infection, seroconversion was likely to be detected amongst them at higher rates than amongst most other groups (incident infections). Sera stored from men who had participated in studies of Hepatitis B incidence and prevalence as well as the efficacy of the Hepatitis B vaccine in the late 1970s presented an ideal opportunity of having ready cohorts in whom seroconversion could be estimated with reasonable precision at a very early stage of the AIDS epidemic in the USA and Europe.
The homogeneity of each group, however, meant that it was difficult to estimate the association between cofactors, such as age at seroconversion, and the IPD. Further, Kaposi’s sarcoma, an AIDS-defrning disease almost exclusively diagnosed in homosexual men, tends to present at levels of immune suppression higher than most other AIDS defining diseases. For this reason it was later recognised that the incubation period for homosexual men may appear to be shorter than for other individuals unless the diagnosis of Kaposi’s sarcoma is ignored (and time is measured to a subsequent disease, if available), or else censored.
31 Table 2.1 summarises findings of the progression estimates from cohorts of homosexual and bisexual men at their longest follow-up times. The median time from seroconversion to AIDS from these studies varies between 8.3 to 10.7 years. That is the time from seroconversion at which 50% of the cohort remain AIDS-free. It can be seen in Figure 2.1 that estimates from these cohorts are in broad general agreement but with the greatest discrepancy 7 to 10 years following seroconversion and with little information beyond 10 years.
2.1.1 UCLA Gay and Lesbian Association
This is a prospeetively gathered cohort of homosexual and bisexual men recruited through UCLA (University of California Los Angeles) Gay and Lesbian Association and a local health club in the spring of 1982 and the summer of 1983. The majority of study participants showed no symptoms indicative of HIV infection at the time of enrolment and none had a frank illness. No-one had developed AIDS within 9 months of enrolment, which the investigators felt was justification for any enrolment bias, which may lead to faster progression rates, being small. Follow-up of participants was nine-monthly.
Taylor et al estimated the date of infection with HIV for each person by using information on the earliest and last possible dates of infection. For persons who seroconverted after enrolment (i.e. ‘incident’ infections) the ‘earliest date’ was the date of the last HIV antibody negative attendance, and the ‘latest date’ was that of the first HIV antibody positive attendance. For persons who were already HIV antibody positive at enrolment (‘prevalent’ infections) the ‘earliest date’ was the latest of three dates: July 1977, the date of the first homosexual sex act, and the date of first living in Los Angeles. The ‘latest date’ for prevalent positive persons was the date of their last visit unless they had presented with AIDS at that visit, in which case it was taken as a date six months earlier.
In total, 177 persons were enrolled of whom 78 were HIV positive (16 incident and 62 prevalent). By March 1986, 9 persons (11.5%) had been diagnosed with AIDS and 15 (19.2%) were lost to follow up. Cumulative AIDS incidence was given as 2% and 11% at 2 and 4 years respectively following HIV seroconversion.
32 Table 2.1
A summary of published studies of HIV progression rates in cohorts of homosexual and bisexual men
Cohort name Main author Country (year of publication) number Cumulative progression in rate to AIDS (and analysis median time) UCLA Gay & Lesbian Association Taylor USA (1986) 78 11% at 4 yrs SFCC* Jason USA (1989) 117 33% at 7.5 yrs SFCC Hessol USA (1989) 135 39% at 9.2 yrs MACS Munoz USA (1989) 1861 21.8% at 5 yrs (10.7 yrs) VLAS Schechter Canada(1989) 102 13% at 5 yrs International Registry of Biggar USA, Canada, Australia, Europe 756 39.8% at 7 yrs Seroconverters* (1990) SFCCRutherford ^ USA (1990) 489 54% at 11.1 yrs VLAS Schechter Canada(1990) 119 36.6% at 6.3 yrs UCLA Gay & Lesbian Association Kuo USA (1991) 81 45% at 8 yrs LA MACS Taylor USA (1991) 908 33% at 7 yrs SFCC Lifson USA (1992) 214 (9.7 yrs) MACS Phair USA (1992) 345 10% at 4 yrs Amsterdam Cohort Hendriks NL (1993) 348 33% at 7 yrs (9.2 yrs) SFCC Buchbinder USA (1994) 588 69% at 14 yrs Tricontinental Study Veugelers NL, Canada, USA, Australia (1994) 403 22.9% at 5 yrs (8.3 yrs) VLAS Hogg"' Canada(1994) 364 55% at 11.5 yrs Hep B cohorts Hessol USA, NL (1994) 362 62% at 12 yrs (10.2 yrs) SEROCO^ Carré France (1994) 443 21.4% at 5 yrs ISS* Rezza Italy (1990) 89 16.2% at 4 years Transfusion Safety Study^ Operskalski USA (1995) 112 36% at 8 years * see also section 2.5.1 t see X see also section 2.5.3 § see also section 2.5.5 U) w Figure 2.1 Published progression rates from HIV seroconversion to AIDS from cohorts of homosexual men
70 -
QU1 o o
40 - o o o o o o o o o o o oc_ o CL o cO □_ 20 - o o
0 - 4 6 8 10 12 14 Time from seroconversion (yrs)
Each symbol represents the AIDS progression estimate (at a given time since seroconversion) given by authors in the publications cited. A cohort may make more than one contribution to the graph.
34 Data censored at the end of June 1989 of 81 persons, of whom 20 had developed AIDS (24.7%) and 38 were lost to follow up (46.9%), gave cumulative progression rates to AIDS at 2 and 4 years following seroconversion similar to those previously reported by Taylor et al and with 45% of subjects developing AIDS 8 years following seroconversion (95% CI= 30%- 60%)
There was no evidence that age at seroconversion was associated with progression to AIDS. The authors noted, however, that almost all the participants were aged 18-30 years.
2.1.2 San Francisco City Clinic Cohort- SFCC
The original cohort was comprised of 6705 men attending the hospital’s sexually transmitted disease clinic who in 1978- 1980 were recruited into studies of the incidence and prevalence of Hepatitis B. In 1980- 1981 a sample of these men (n=359; 34 from the Hepatitis B Prevalence Study and 325 from the Hepatitis B Incidence Study) were recruited into a clinical trial of the efficacy of a plasma derived Hepatitis B vaccine.
Starting in October 1983, investigators attempted to contact all the original 6705 men with the view of setting up an HIV cohort through the testing of sera left over from the Hepatitis B studies. 4043 men were still alive (of whom 2877 consented to their bloods being tested), 699 had died (for 675 of these stored bloods were tested), and 1963 were lost to follow up (and whose blood was not tested for HIV).
Of the 2877 men alive who consented to their bloods being tested 1431 (49.7%) were HIV negative and 1446 (50.3%) were HIV positive (223 prevalent (i.e. positive on all sera tested), 165 incident with an HIV test interval of 24 months or less, and 1058 incident with a test interval greater than 24 months).
Of the 675 who had died whose bloods had been tested, 143 (23%) were HIV positive (130 prevalent and 23 incident). Most publications on data from this cohort is derived from the prevalent living subjects, the incident living subjects with a test interval of 24 months or less, and a subset of seroconverters from those who had died.
35 The date of seroconversion was estimated as the mid-point between the last negative and first positive test dates for incident seroconverters, and was determined for those with prevalent infection using a probability estimate based on the men with negative dates.
Hessol et al analysed data on the subset of men who had been enrolled in the Hepatitis B vaccine trial in 1980-1. Of 359 men, 320 agreed to their stored bloods to be tested for HIV. Of these, 158 were HIV positive (20 prevalent and 138 seroincident). The 39 persons whose sera were not tested for HIV had either declined (n= 22) or had become lost to follow up (n= 17).
From data on 135 men who were either incident seroconverters and with a window of less than 25 months between test dates (n=117) or prevalent positive (n=18) censored in mid-May 1989, the authors estimated the cumulative rate of progression to AIDS 9.2 years after seroconversion to be 39% (95% CI= 27%- 51%), slightly lower than estimates given by UCLA study
There was no significant effect of age at seroconversion, race or calendar time of seroconversion on the rate of progression but as the cohort was fairly homogeneous with respect to those factors the ability to detect such differences may have been limited.
Three further publications reporting the analysis of data on men fi*om the larger Hepatitis B cohort gave faster progression estimates than that for the men in the vaccine trial.
Data on 489 men censored at the end of December 1989, analysed by Rutherford et al gave estimates of cumulative progression to AIDS at 5, 10, and 11.1 years following seroconversion of 13%, 51%, and 54% respectively.
Lifson et al analysed data on 214 men with laboratory measurements for CD4 cell counts, p-2 microglobulin, and p24 antigen levels available and who were without AIDS at the time of these laboratory measurements censored at the end of November 1991. The authors gave an estimate of median time to AIDS of 116 months (9.7 years).
36 Data censored at the end of December 1992 on 588 men gave estimates of progression to AIDS 3, 10 and 14 years following seroconversion as 3%, 51% and 69% respectively^^.
2.1.3 Amsterdam cohort of homosexual men
This prospective cohort was set up in October 1984 and initially enrolled sexually active homosexual men who were community volunteers. Almost one third of these men were HIV positive when enrolled, a number of whom had been participating in a Hepatitis B vaccine efficacy trial and specimens of stored bloods were still available. From February 1985 the cohort started enrolling HIV negative men only. Follow up was 3-6 monthly. By the end of February 1990, 1004 men had been enrolled in total of whom 269 were already positive. Further, 79 men seroconverted while under follow up.
Analysis of data on these 348 positive men was censored at 1 February 1990 The dates of seroconversion for the 269 prevalent positive men were estimated using a probability density function for the expansion of the observed part of the incubation assuming different distributions. The authors’ preferred estimate, which allowed for a slowing of the hazard rate at 7 years following seroconversion, gave a median time to AIDS of 9.2 years.
By October 1991, 108 men had seroconverted under follow up, of whom 28 had developed AIDS Keet et al evaluated the role of clinical symptoms and laboratory markers present during primary HIV infection on progression estimates and found that: the presence of fever and skin rash, absence of antibodies to HIV core proteins and having detectable p24 antigen at seroconversion were all independent predictors of progression to AIDS. The authors concluded that even at a very early stage of HIV infection it is possible to predict persons at high risk of rapid disease progression.
2.1.4 Other Hepatitis B vaccine cohorts
Hessol et al reported on an analysis of pooled data from 362 homosexual men with well documented times of seroconversion from three trials of Hepatitis B vaccine efficacy.
37 The men in the HIV study had consented to have their bloods tested for HIV in the early 1980s, some of whom were found to be prevalent HIV positive and others were with incident infection. The studies were in Amsterdam (n=74), New York City (n=120) and San Francisco (n=168).
Data were censored in 31 December 1991 with all those not known to have died being censored at that date. For the analysis of time to AIDS, New York men were censored at the date of last visit, the median time from seroconversion to the development of AIDS was estimated to be 10.2 years with a cumulative progression rate of 62% 12 years following seroconversion (95% CI= 53- 70%). More recent calendar time was associated with faster progression to AIDS with those observed in the period July 1989 to December 1991 having a relative hazard of 1.77 (95% CI= 1.01- 3.10) relative to the time period before 1987.
A slight increase in risk was observed with each 10-year increase in age at seroconversion (RH per 10 year increase = 1.31, 95% CI= 0.95- 1.80) though this was not statistically significant.
The median time from seroconversion to death was estimated to be 12.3 years (95% CI= 10.75- 12.75 years). Older age at seroconversion was associated with faster progression to death with a relative hazard of 1.65 (95% CI= 1.12- 2.41) for each 10 year increase in age. No evidence of a change in calendar time was observed, however, leading the authors to conclude that the expansion of the AIDS case definition in 1987, improved diagnostic methods and a lower diagnostic threshold by clinicians have resulted in an AIDS diagnosis being made at an earlier stage of HIV infection in more recent time periods.
2.1.5 Vancouver Lymphadenopathy AIDS Study -VLAS
This is a prospective study of 715 homosexual men recruited from 6 general practices in central Vancouver during the period November 1982 to February 1984. Until September 1986 follow up of subjects was 6-monthly, thereafter it became annually.
38 Schechter et al reported on analyses of data on 348 men (246 prevalent positive and 102 ineident), who completed at least 2 visits during the period November 1982- November 1987, and censored in July 1988. Progression rates given are from the first antibody positive test for those seroprevalent (and will therefore under-estimate the time from seroconversion to AIDS) and from the mid-point of the negative and positive test dates for the sero-incident members.
Five years following seroconversion (for incident seroconverters), and following the diagnosis of HIV (for prevalent positive subjects), 23% (95% CI= 17.5- 28.5%) of prevalent positive subjects progressed to AIDS, compared to 13% (95% CI= 4.6- 21.4%) of those with incident infection. The factors found to be independent predictors of AIDS were: an elevated IgA level (> 20 mg/dl), low CD4 counts (< 400 cells /ml), and having more than 20 male partners from a high risk area (San Franeisco, Los Angeles, New York) in the 5 years previous to enrolment.
Schechter et al further reported on 18 men with incident HIV infection who had progressed to AIDS by July 1989 and were treated as ‘cases’ each with 3 randomly selected controls who had seroconverted within 3 months of the case but who had not progressed to AIDS .
They concluded that abnormalities which are predietive of more rapid progression to AIDS appear early in HIV infection and estimated a progression rate to AIDS of 36.6% (95% CI= 21.6- 51.6%) 76 months (6.3 years) after HIV seroconversion.
Hogg et al reported on 364 HIV positive men censored in December 1993 (234 prevalent and 130 incident infections). The cumulative progression to AIDS 11.5 years after seroconversion was estimated to be 55%. Progression rate to what was considered to be HIV-related death was estimated to be 45% at 11.5 years after seroconversion.
2.1.6 Multicenter AIDS Cohort Study- MACS
This is a prospective cohort of 4954 homosexual men without AIDS at enrolment who were followed up on a six-monthly basis in 4 cities in the USA (Los Angeles, Chicago,
39 Baltimore and Pittsburgh) since mid-1984. Of these, 1745 were HIV antibody positive at enrolment.
In the first 4 years after enrolment, 268 men seroconverted to HIV. Munoz et al reported on findings from data on 1861 men (1628 seroprevalent with values on baseline haematological variables available, and 233 sero-incident with a maximum of 8 months between the negative and positive test dates). Seroconversion was estimated as the mid-point between antibody tests for those with incident infection, and markers of infection from the seroincident members of the cohort were used to estimate the time from infection for those seroprevalent.
Progression to AIDS 5 years after seroconversion was estimated to be 21.8% (95% CI= 16-25%), with a median estimate at 10.7 years (95% CI= 8.4-19.3 years).
Data on 345 men who seroconverted by July 1989 were analysed, of whom 32 (9.3%) developed AIDS by March 1990 Estimated progression rates to AIDS, of 6% and 10% 3 and 4 years following seroconversion respectively, were much lower than those made by Munoz suggesting that either the time of seroconversion for the prevalent members of the cohort (not included in analysis by Phair et al) was estimated to be later than the actual date or the IPD had lengthened.
Taylor et al reported on data from 908 men enrolled in the Los Angeles portion of the MACS study between April 1984 and February 1985 (99 seroconverters and 809 prevalent positive persons). Data available in January 1990 were censored in July 1989. The authors reported on a slight (statistically non-significant) lengthening of the incubation period with 28% of persons developing AIDS within 6 years of seroconversion for those infected in the first half of 1979 compared with 25% of persons infected in the first half of 1983. Assuming no secular changes in the IPD, 27% (95% confidence intervals = 23, 31%) are estimated to have developed AIDS within 6 years of HIV seroconversion and 33% within 7 years.
40 2.1.7 Tricontinental Seroconverter Study
Data on HIV positive homosexual and bisexual men who had seroconverted under follow up were pooled from the following 5 cohorts:
1. VLAS- Canada. 128 seroconverters from a group of 715 men recruited from 6 general medical practices in November 1982- December 1984. 2. SAPS- Australia. 74 seroconverters from a group of 1057 men recruited from private clinical practices in 1984-1985. 3. Amsterdam cohort study- Netherlands, 137 seroconverters from a group of 748 men, community volunteers, who were enrolled before February 1985 (thereafter only HIV negative men were enrolled). 4. San Francisco General Hospital (SFGHS)- USA. 19 seroconverters from a group of 462 men enrolled from 1983. 5. San Francisco Men’s Health study (SFMHS)- USA. 45 seroconverters from a group of 807 men.
Data on all 403 seroconverters were analysed By November 1992, 114 men were diagnosed with AIDS (28%), and 67 died. The mean age at seroconversion varied with the lowest (31 years) in Vancouver cohort and highest (35 years) in the San Francisco cohorts.
Progression to AIDS 3 and 5 years after seroconversion were estimated to be 6.2% (95% CI= 3.9- 9.1) and 22.9% (95% CI= 18.0- 27.9) respectively, and a median time of 8.3 years (95% Cl =7.4- 9.1) was estimated. This represents the shortest published median time to AIDS. No significant differences were found between the cohorts with respect to progression times to AIDS, and no age effect was found (RH per 10-year increase = 1.11, 95% CI= 0.85-1.46). Hazard rates to AIDS increased slightly over time to 1986 but were not statistically significant.
Cumulative progression to death 5 years after seroconversion was estimated to be 8.7% (95% CI= 5.4- 12.0), with a median of 8.9 years (95% CI= 8.7- 9.3). No difference was found between the cohorts with respect to rates of death. Risk of death was higher for
41 those who seroconverted in 1985 compared to those who seroconverted before 1984 (RH= 2.36, 95% CI= 1.00- 5.54). Use of prophylaxis was associated with a decreased risk whereas the use of AZT was associated with an increased risk, neither was found to be statistically significant however. Older age at seroconversion was associated with faster progression to death with a relative hazard of 1.50 (95% CI= 1.06- 2.11) for each 10-year increase in age.
2.1.8 Homosexual men in Sweden
All records of HIV infected homosexual men who had ever attended any of 3 participating clinical centres were reviewed including those who had been diagnosed with AIDS, became lost to follow up or had died. The medical records of all 389 eligible men were retrospectively examined in mid- 1988. For 102 men the date of seroconversion could be estimated within a 12-month period either from the availability of a previous antibody negative test, from the patient’s history including the results of contact tracing, or from evidence of primary HIV infection.
By mid- 1988, 7 of the 102 seroconverters had developed AIDS giving a mean time to the development of AIDS of 97 months (8.1 years)
2.2 COHORTS OF HAEMOPHILIACS
Of great advantage to these cohorts of infected haemophiliacs is that the rate of loss to follow up is much lower than for other groups of infected individuals as they are undergoing clinical follow up for their clotting disorders in any case. As follow up is more complete than for cohorts of homosexual men, this may lead to estimates of longer time to AIDS than for homosexual men (see Section 4.6.1).
In fact, progression estimates overall tend to be slower in haemophilia cohorts (i.e. the IPD is longer) than those estimated from cohorts of homosexual and bisexual men. This may, in part, be due to the younger age profile of the haemophilia cohort but also the almost absence of the diagnosis of Kaposi’s sarcoma in this group of individuals.
42 Furthermore, the high rate of mortality among these individuals, for reasons associated with their clotting disorder, leads to slower estimates of progression to AIDS as those who die without the diagnosis of AIDS are censored at the time of death.
All haemophilia cohorts found a strong association between age at seroconversion and progression to AIDS and to death, with older persons having higher rates than younger individuals.
Table 2.2 summarises findings of the progression estimates from cohorts of haemophiliacs at their longest follow-up times. These are not directly comparable, however, as the progression rates given for the whole cohort will very much depend on its age profile.
2.2.1 UK Haemophilia cohorts
The UK Haemophilia Centre Directors’ Organisation (UKHCDO) cohort of 1201 haemophiliacs with HIV alive on 1st January 1985 was retrospectively assembled in 1986. Stored sera were tested to establish when seroconversion was likely to have taken place. For 341 persons a sample was found to be negative to HIV antibodies. For those positive at the earliest available sample it was assumed that they were negative on 1 January 1979. In order to estimate the date of seroconversion for the whole cohort (regardless of the availability of negative sera) the authors assumed that no-one was positive before 1 January 1979 and that the probabilities of seroconverting per unit time within each of 3 defined calendar time periods from 1 January 1979 and 31 December 1987 were calculated according to a ratio (suggested from an initial analysis of the data) in which the cumulative distribution of infections was estimated using methods for interval censored data
At the end of December 1987, 47 persons had died without AIDS (and were censored at the date of death) Censored in December 1987, the rate of progression to AIDS was found to be strongly associated with age - persons aged over 45 years having a 5-fold risk of developing AIDS 5 years after seroconversion compared to persons aged under 25 years (Table 2.3). Further, those with an explicit negative test were found to have a higher risk of AIDS. This may be an indication that their imputed negative dates
43 Table 2.2 A summary of published studies of HIV progression rates in cohorts of persons with haemophilia
Main author Country number in Cumulative progression (year of publication) analysis rate to AIDS Eyster USA (1987) 92 18% - 6 yrs Giesecke Sweden (1988) 98 5% - 5 yrs Darby UK (1989) I20I 7% - 5 yrs Jason USA (1989) 79 27%- 7.5 yrs Goedert USA (1989) 319 25.1% - 9 yrs Darby UK (1990) 1201 13% at 7 yrs Biggar USA, Canada, Australia, 297 27.3 % - 7 yrs (adults) Europe (1990) 19.7% - 7 yrs (children) Ragni USA (1990) 84 49% - 7.5 yrs Schinaia Italy (1991) 499 12.8% - 7 yrs Phillips UK (1991) 63 (>30 yrs: 7 yrs) Lee*^ UK (1991) 111 45% at 11 yrs Chiarotti Italy (1994) 732 14.9-17.8%-8 yrs Operskalski USA (1995) 519 24% at 8 yrs
j see also section 2.5.4 i see also section 2.5.1 § see also section 2.5.5 were too early, and that such persons may have seroconverted at a later time. An analysis of 63 persons with negative and positive dates for HIV available through stored sera (range 15 days- 24 months between tests) censored in November 1989 gave estimates of progression to AIDS at 7 years after seroconversion to be 50% for those aged over 30 years, and 12% for those aged 10-19 years These are part of a cohort of 111 HIV positive haemophiliacs attending a single haemophilia centre in the UK with a median age at first positive test of 24 years (range 2-77 years). Analysis censored at the end of November 1990 gave estimates of progressing to AIDS 11 years after seroconversion as 45% (95% confidence intervals = 27, 57%), with 31% (95% confidence intervals =12, 50%), and 56% (95% confidence intervals = 39, 73%), progression rates for those aged under 25 years and 25 years and more respectively These estimates are somewhat faster than those made by Darby et al for the UK group as a whole
2.2.2 Multicenter Haemophilia Cohort Study- MHCS
Data were collected beginning in 1985 on 1219 haemophiliacs attending 11 centres in USA. Of 674 HIV infected haemophiliacs, data on 319 persons for whom a serum specimen stored tested negative for HIV antibodies were censored in November 1988. The overall estimate of progression to AIDS at 9 years post seroconversion was given as 25.1% (95% CI= 20.3- 29.9). A strong association was found between the risk of progressing and age at seroconversion (Table 2.4).
2.2.3 Pennsylvania cohorts
Data were retrospectively collected on 111 haemophiliacs at a haemophilia centre in central Pennsylvania, USA, enrolled in September 1982. Data on 92 HIV positive persons (including 9 incident infections) were analysed, 9 of whom had progressed to AIDS and 8 died (2 were without AIDS) by September 1986. Cumulative progression to AIDS or AIDS-related illness (one person) was estimated to be 18% at 6 years after seroconversion. Older age at seroconversion was significantly associated with a higher incidence of AIDS (2-21 years of age versus 22-64 years)
45 Table 2.3 Estimates of progression to AIDS by age group at seroconversion from the UKHCDO data
Time from Percent progressing to AIDS seroconversion Age at seroconversion (yrs)
(years) <25 25-44 45+ All ages
3 1 2 9 3 5 4 6 19 7 7 6 20 34 13
Table 2.4 Estimates of progression to AIDS within 9 years of seroconversion by age group for persons in the MCHS 82
Age group Percent with AIDS 9 years after (yrs) seroconversion (± 1 s.e.)
1-17 13.3 (±7.7)
18-34 26.8 (± 8.3)
35-70 43.7 (±16.4)
Overall 25.1 (± 7.0)
46 Another cohort of haemophiliacs, part of the MCHS, from the Hemophilia Center of Western Pennsylvania consisted of 84 patients all with a negative HIV antibody test followed by a positive test within 36 months Of these, 30 (36%) had developed AIDS by 1 May 1989, and none had developed AIDS within 2 years of seroconversion, estimated as the mid-point between the last negative and first positive antibody tests. The proportion developing AIDS within 4, 6 and 7.5 years following seroconversion was estimated to be 12± 4%, 28± 5% and 49± 8% (± SE) respectively. A strong age effect was also observed with the proportion developing AIDS by 6 years following seroconversion for those aged over 30 years, 18- 30 years, and under 18 years of age being 49± 10%, 24± 9% and 5± 5% (± SE) respectively.
2.2.4 Italian Haemophilia cohort
Data on HIV positive haemophiliacs from a national haemophilia registry established in Italy in 1988 were analysed. To the end of December 1989, 499 HIV positive haemophiliacs were reported without other risk factors for HIV infection. Dates of seroconversion were estimated in the same manner as Darby et al for the UKHCDO Seven years after seroconversion, 12.8% progressed to AIDS. Significantly faster progression rates were seen in persons aged 35 years or more (27% compared with 14.5% for those aged 12 years or less, and 8.9% for 13-34 year olds)
Using data censored at the end of June 1991 Chiarotti et al fitted different models assuming a number of parametric distributions in order to estimate both seroconversion time and the incubation distribution Of 732 persons analysed, 118 had developed AIDS (16%), 586 were alive without AIDS (80%), and 28 died without AIDS (4%) by the end of June 1991.
Estimates 8 years after seroconversion ranged from 14.9- 17.8% using the Kaplan-Meier method, 14.1-17.2% using a Weibull probability density function, 14.5-17.3% using a Generalised Exponential model, and 14.4-17.3% using a log-logistic model, the range given depending on the assumed distribution for estimating the time of seroconversion.
47 2.3 COHORTS OF TRANSFUSION RECIPIENTS
Transfusion recipients presented early in the AIDS epidemic as the most obvious group of infected individuals for studies of the incubation period as the date of infection was much better defined than in other groups. However, because they tend to be ascertained as infected through illness and particularly through the diagnosis of AIDS, estimates derived from them are akin to estimating survival only for persons who have died (Section 4.5). These data provided no information on persons who have not developed AIDS and cases with longer reporting lag are excluded as are persons who had died soon after the transfusion. Those unaware of their infection, usually because they are asymptomatic and/or are at an earlier stage of disease, are generally excluded. All these factors will lead to short estimates of the IPD. In order to overcome this some investigators have attempted to ascertain transfusion recipients through Took back’ exercises with the aim of including all those eligible, regardless of clinical state. These cohorts are summarised in sections 2.3.4, 2.3.5, and 2.3.6.
Further, estimates of the IPD derived from transfused persons are generally much shorter than for other groups due, not only to their older age profile, but also for the reasons underlying a medical condition which had necessitated the transfusion in the first instance. Estimates of the IPD are, therefore, bound up with estimates of ‘background’ survival.
Table 2.5 summarises findings of the progression estimates from cohorts of transfusion recipients at their longest follow-up times. As with haemophilia cohorts, progression estimates from different studies are not directly comparable as they will depend largely on the age profile of the cohort.
2.3.1 Transfusion-associated AIDS cases reported to CDC
These data provided the first estimates of the AIDS incubation period distribution. 83 transfusion-assoeiated AIDS cases with known dates of transfusion diagnosed by December 1984 and reported by April 1985 to CDC, Atlanta, were analysed The simple mean incubation period was estimated to be 2.6 years. The authors, however.
48 Table 2.5 A summary of published studies of blood transfusion recipients
Main author Country number in Cumulative progression (year of publication) analysis rate to AIDS (and median time)
Lm'" USA (1986) 83 (4.3 yrs) Medley USA (1987) 297 (0-4 yr olds: 1.9 yrs 5-59 yr olds: 7.79 yrs 60+ yr olds: 5.44 yrs) Giesecke Sweden (1988) 48 29% - 5 yrs Ward USA (1989) 101 49.2% at 7 yrs Blaxhult Sweden (1990) 58 34% at 5 yrs Msellati France (1990) 87 34.2% - 5 yrs (6.08 yrs) Downs Europe (1991) 488 (14 yrs'or less: 2.3 yrs, 15 yrs or over: 9.8 yrs) Kopec- Australia (1993) 117 (7.2 years) Schrader see also section 2.5.4 4^ VO recognising that cases with long ineubation period would not yet be diagnosed and henee have an incubation period exceeding the truncation period, attempted to correct for this through modelling. Assuming a parametric distribution they gave an estimate of the mean incubation period of 4.5 years (90% CI= 2.6- 14.2 yrs). Medley et al later examined data from 494 transfusion-associated AIDS cases diagnosed in the US between January 1982 and June 1986 and reported to CDC by January 1987. Of the 297 cases with known dates of transfusion, 135 (45%) were aged 60 years or over. Women were found to have a lower, although not statistically significant, AIDS risk than men, with an estimated median of 8.36 years for women and 5.5 years for men. Persons aged under 4 years of age at infection were found to have the shortest time to AIDS, but thereafter the hazard increased with age.
2.3.2 Cohort Aquitaine
Data were collected in 1987 retrospectively, and thereafter prospectively on adults who had been infeeted through reeeipt of HIV positive blood in Bordeaux, France This is part of a much larger cohort of HIV infected persons with defined time of infeetion, normally the availability of a previous antibody negative test.
In total, 89 persons transfused between August 1981 and June 1985 and diagnosed between 1985 and 1989, one third of whom had already been diagnosed with AIDS were identified. Analyses were censored at the end of June 1990. The mean age at transfusion was 38.7 years, median 36 years (range 14-81 yrs). 56 persons (63%) had been tested because they presented with signs and symptoms compatible with HIV infection.
The authors estimated the median time to AIDS to be 73 months, with progression rates of 8% (95% CI= 3.8- 16.2) and 34.2% (95% CI= 20.3- 49.3) 3 and 5 years after transfusion, respectively. Survival at 5 years post transfusion was estimated to be 81.7% (95% CI= 72.5- 90). Persons aged 40 years or over and women were found to have higher rates of progression. The authors also reported that the larger the inoculum (volume of infected blood transfused) the higher the risk for progression was likely to be.
50 2.3.3 European Transfusion-Associated AIDS Data Set- ETAADS
National AIDS surveillance data from 32 countries reported to the WHO collaborating centre for the European region in Paris by the end of 1989 and diagnosed by the end of 1988 with known dates of transfusion were analysed The data analysed consisted of 51 AIDS cases in children (aged under 14 years), and 437 AIDS cases in adults (aged 15 years or over) transfused between 1978 and 1988. Parametric maximum likelihood estimations were made assuming a parametric form of the distribution of the incubation period, with the resulting likelihood function being conditional on the diagnosis of AIDS. Further, estimates of the annual incidence from 1978 to 1985 of infective transfusions expected to give rise to AIDS cases were obtained from the number of cases already diagnosed per month of transfusion. In this manner the authors attempted to account for transfusion- associated HIV infections not yet diagnosed.
The authors estimated the median time from seroconversion to AIDS to be 9.8 years for adults and 2.3 years for children.
2.3.4 Transfusion-associated cases in California
The authors initially ascertained all known and possible donors of HIV infected blood through matching with local AIDS registries in California. They subsequently reviewed records of recipients and attempted to enrol all persons who had received infected blood during the period January 1979 and August 1985 into the study Any recipient identified whose risk for HIV infection was identified through the diagnosis of AIDS was excluded from the analysis in order to minimise ascertainment bias. In total 765 recipients were identified, 694 of whom had records available.
Of the 694 persons with available records, 437 had died (16 were found to have been HIV infected, 13 with AIDS, one was antibody negative, and the serostatus of 420 were unknown). Of the 257 persons with records who were still alive, 119 were HIV positive (6 of whom had AIDS and were therefore excluded from analysis), 84 were antibody negative, and the serostatus of 54 was unknown. Of the 113 HIV positive recipients without AIDS, 82 were enrolled into the study and their data analysed (70 had other risk factors for HIV, 10 had moved from the area, and 14 declined participation).
51 The authors found that 49% of recipients from donors in whom AIDS developed soon after the donation progressed to AIDS 4 years after infection, compared with 4% in the other group. Estimates of progression rates to AIDS after 1, 5 and 7 years from seroconversion are given as 0.5, 32.6, and 49.2% respectively.
The authors further reported that inoculum size and virus strain variation may be important factors in progression.
2.3.5 Swedish Transfusion recipients
The authors conducted a ‘lookback’ exercise of transfusion recipients in Sweden through tracing the recipients of known infected blood (55), through general screening (2), and through contact tracing (1). Of the 58 persons found to have been infected via a transfusion 29 were men and 29 women (over 15 years of age). Analyses censored in July 1989 found a strong age association, with the hazard increasing as age increases Table 2.6 gives cumulative progression estimates for those under 60 years of age and those aged 60 years or more.
The authors also found a tendency towards slower progression for females which was not statistically significant.
2.3.6 Sydney Red Cross Cohort
In July 1984, a registry was established in Sydney, Australia to collect information on all persons infected with HIV through receipt of a blood transfusion. Notification to this system was through two methods: the independent diagnosis of HIV infection in such persons, and also through a ‘look-back’ program of recipients of blood units whose donors were found to have been infected at the time of the donation. All persons who had received a transfusion in the period January 1980 to April 1985 were included on condition of having no other risk factors for HIV infection. The cohort was largely retrospective in nature.
52 Table 2.6 Progression to AIDS by age group at seroconversion for Swedish transfusion recipients 90
Years from Age when transfused (years) seroconversion (years) 15-59 60+ All ages
1 0 0 0 2 4 3 4 3 16 19 20 4 16 23 22 5 20 39 34
53 By 31 December 1991, a total of 117 cases of transfusion-associated HIV infection with a date of infection that could be estimated had been reported. Survival times were censored at date last seen for those last seen before 31 December 1991
To the end of December 1991, 67 persons (57%) had developed AIDS. The authors estimated a median time from infection to AIDS to be 7.2 years (95% CI= 5.4- 9.3). No significant difference in progression to AIDS rates was found between men and women. A slightly increased risk was associated with older age, with median progression times of 7.8 years (95% CI= 4.5- 10.1), 7.1 years (4.4- 9.5) and 5.4 years (2.9- 8) for the age groups 13- 39, 40- 59 and over 59 years respectively. If younger persons were more likely to have been lost to follow up, however, this would result in a dampening of the age effect.
The authors further reported that a significantly greater proportion of individuals who had never received AZT (treated as a time- dependent variable) developed AIDS compared to individuals who had (p= 0.04).
2.4 COHORTS OF INJECTING DRUG USERS
Few studies have reported on follow up of injecting drug users from seroconversion. Like persons with haemophilia progression estimates overall tend to be slower than those estimated from cohorts of homosexual and bisexual men due to their younger age profile, the absence of Kaposi’s sarcoma, and the high rate of mortality in infected persons who do not develop AIDS. Further, rate of loss to follow up tends to be higher than in other groups.
Table 2.7 and Figure 2.2 summarise the findings of progression estimates from cohorts of injecting drug users.
2.4.1 Italian Seroconversion Study- ISS
This multi-centre prospective cohort was set up in late 1987 as a collaboration among a number of clinical centres caring for HIV infected persons in Italy. Originally the cohort included only injecting drug users having a documented negative test for HIV
54 Table 2.7 A summary of published studies of injecting drug users
Cohort name Main Country no. in Cumulative progression rate author (year of publication) analysis to AIDS Italian Seroconversion Study Rezza Italy (1989) 205 17.9% at 4 years* Italian Seroconversion Study| Rezza Italy (1990) 261 13.8% at 4 years Italian Seroconversion Study Anon Italy (1992) 468 21.2% at 7 years Edinburgh City Hospital Cohort Flegg"" UK (1994) 283 19% at 8 years
* progression to AIDS or death
t see also section 2.5.3
LA LA Figure 2.2 Published progression rates to AIDS from cohorts of injecting drug users
20 -
QLO
oC 10 -ML o CL cO CL
0 -
3 4 5 5 Time from seroconversion (yrs)
Each symbol represents the AIDS progression estimate (at a given time since seroconversion) given by authors in the publications cited. A cohort may make more than one contribution to the graph.
56 followed by a positive one within a period of 18 months. The protocol later expanded to include persons infected by other means and to widen the window between negative and positive test dates to 2 years. A number of participating centres had already set up mechanisms for identifying eligible seroconverters and data from these were retrospectively included in the ISS. Follow up is six monthly, and in order to reduce possible loss to follow up of subjects, data were linked with those from the national AIDS registry. The vital status of persons lost to follow up is cross checked against records at the Municipality of birth of participants. In this manner, after allowing for sufficient time for reporting delay to the AIDS and death registries, the investigators feel they can reliably regard persons not found on these registries to be event-free at the analysis cut-off date. Analysis of data censored in April 1988 of 205 injecting drug users who seroconverted between 1980 and 1988 produced progression estimates to AIDS or to death without AIDS of 1.5, 2.9, 9.9 and 17.9% in the first, second, third and fourth years following seroconversion respectively
In a later publication, data on 468 IDUs were analysed and persons who died without an AIDS diagnosis were censored The authors estimated the risk of developing AIDS to be 21.2% (95% CI= 9.4- 33.0%) within 7 years following seroconversion. Persons aged 25 years or over were found to progress much faster than persons aged less than 25 years {p= 0.007). No evidence of a difference in progression rates was found between men and women.
2.4.2 Edinburgh City Hospital Cohort- ECHC
This retrospective cohort was set up at the beginning of 1985 of persons attending a single infectious disease clinic in Edinburgh, Scotland. Sera had been stored on a large number of clinic attenders from the mid 1970s for studies on Hepatitis B following a large nosocomial outbreak which had occurred in 1970 in the city. Only sera of persons who were still attending the clinic were tested and in this manner it was possible to establish the dates of seroconversion for a number of them dating back to 1983. The date of seroconversion could be estimated either because a specimen testing negative was found or because the earliest specimen tested positive and was within 2 years of 1 January 1983 (the earliest estimated possible time of the start of the HIV epidemic in
57 Edinburgh). The cohort consists mainly of injecting drug users, but also includes a number of homosexual men and persons infected heterosexually.
Data on 403 HIV infected IDUs reviewed in the unit by January 1992 were presented by Flegg et al Of these 403 individuals, 348 had accurately estimated dates of seroconversion, of whom 283 were asymptomatic at the initial presentation. The analysis of progression rates concentrated on this sub-group of asymptomatic patients. Of these, 177 were male (62.5%).
Persons judged not to be under regular follow up were censored at the date they were last seen in the clinic. By January 1992, 37 (13.1%) had been diagnosed with AIDS and 40 (14.1%) had died, of whom 23 had not been diagnosed with AIDS. Progression estimates to AIDS 5, 6, 7 and 8 years following seroconversion were given as 2, 7, 13 and 19% respectively. No evidence of an age effect was observed.
The slow progression estimates resulting from this analysis are most likely to be due to the preferential inclusion of long term survivors (because those who died before the cohort was set up were not included) and the exclusion of persons with clinically diagnosed illness at the initial presentation, as they are likely to have more rapid disease progression.
2.5 STUDIES COMPRISING DIFFERENT GROUPS
2.5.1 Homosexual men and men with haemophilia
Jason et al compared AIDS progression rates in 79 HIV infected men with haemophilia from a haemophilia treatment centre in western Pennsylvania to that of 117 homosexual and bisexual men enrolled in the SFCC and with a maximum HIV test interval of 24 months. Censoring data at the end of December 1988, no significant differences were found between the two groups in terms of progression to AIDS, although progression rates among men with haemophilia appeared to be slightly slower than that among homosexual and bisexual men (Table 2.8)
58 An International Registry of Seroconverters (1RS) was compiled in October 1987 to pool available data from known cohorts of HIV infected persons world-wide in order to examine the influence of a number of cofactors on progression to AIDS, particularly: exposure category, age, calendar time of infection and geographical area Data on 1171 subjects with an HIV test interval of 25 months maximum and enrolled in prospective studies were analysed. Follow-up was typically 6-monthly, with no follow- up after 1987 (so as to exclude the possible effect of treatment and prophylaxis). Persons event-free were censored at the time last seen and death due to HIV was treated as an event. The data consisted of: 756 homosexual and bisexual (median age 36 yrs), 297 haemophiliacs (median age 31 years adults, and 12 years for children), 82 injecting drug users, 22 heterosexuals without other risk factors, and 14 transfusion recipients.
Homosexual and bisexual men were found to progress more rapidly than haemophiliacs, mainly due to the diagnosis of Kaposi’s sarcoma in homosexual and bisexual men.
Children with haemophilia were found to progress slower than adults with haemophilia. Beyond childhood, however, the age effect was confusing: 18-29 year olds had similar progression rates to 30-44 year olds but slower rates than those aged 45 years and over. No such observation in homosexual and bisexual men was made, however. The authors concluded that this could be due to older haemophiliacs being more likely to be diagnosed with CMV disease and chronic liver failure presenting later in life and thus accentuating the age effect.
Using pooled data on 1020 homosexual men from the 1RS and 373 HIV-positive haemophiliacs from the MHCS Rosenberget al further examined the effect of age at seroconversion on the incubation period distribution Follow up was censored at 1 July 1987 for both cohorts. After adjusting for age in a non-parametric proportional hazards model the authors found progression rates were 2.12 fold higher among homosexual men than haemophiliacs (95% confidence intervals = 1.36, 3.30) and this, in part at least, was found to be due to homosexual men having a greater risk of developing KS as an AIDS-defining condition. The linear effect of age for haemophiliacs was found to be 1.049 per year (95% confidence intervals = 1.026- 1.074) equivalent to a relative hazard of 1.6 per 10-year increase in age. For homosexual men the point estimate for a one year increase in age was 1.031 (95%
59 confidence intervals = 0.99- 1.07) corresponding to a 1.4-fold increase in the hazard for each 10-year increase in age.
Biggar et al found no evidence of a change in the IPD over time, and persons with a period greater than 25 months between negative and positive tests were found to have a higher risk of progression (hence their exclusion from the main analyses). Estimates of cumulative progression rates to AIDS found by Biggar and colleague were similar to those made by Jason et al (Table 2.9).
2.5.2 Homosexual and heterosexual men
Data on persons reported to the SEROCO study were analysed. SEROCO is a prospective study of HIV positive adults enrolled at 18 centres in France. Follow up of subjects is 6-monthly and information on any previous antibody negative test to HIV is also collected. Haemophiliacs are excluded from the study as they are followed up by another group of investigators.
Of 599 subjects who had previously tested negative for HIV, 443 persons (312 homosexual and bisexual men, and 131 heterosexual men) with dates of seroconversion were included in this analysis and censored in April 1993 377 had a negative test date within 2 years of the first positive one. Of 47 homosexual men diagnosed with AIDS, 19 were diagnosed with KS. Median age at seroconversion was 29 years (range 19-61 years) for homosexual and bisexual men, and 25 years (range 16-61 years) for heterosexuals. Progression to AIDS was more rapid in persons aged 40 years or more at seroconversion compared to those under 40 years. The risk of AIDS at 5 years following seroconversion was estimated to be 6.9%, 15.5%, and 42.2% for those aged under 20 years, 20-39 years, and 40 years or more respectively. The relative risk per 10 year increase in age was 1.34 (95% CI= 1.03, 1.77).
The risk of AIDS at 5 years after seroconversion was estimated to be 7.9% for heterosexuals and 21.4% for homosexual and bisexual men. The relative risk of progression to AIDS for homosexual and bisexual men compared to heterosexuals =2.42 (95% CI= 1.2-4.9) after adjusting for the effect of age.
60 Table 2.8 A comparison of progression rates to AIDS between haemophiliacs in Western Pennsylvania and homosexual men in the SFCC 96
Time from Percent with AIDS seroconversion (95% confidence intervals)
(yrs) Haemophiliacs Homosexual and bisexual men
1.5 0 0 2.5 1(0- 4) 5 (1- 9) 3.5 10(3.5- 16.5) 11 (5- 17) 4.5 12(5- 19) 16 (9- 23) 5.5 16(7.5- 24.5) 24(16- 32) 6.5 24(13.5- 34.5) 3 0 (2 1 -2 9 ) 7.5 27(15- 39) 33 (23- 43)
Table 2.9
A comparison of progression rates to AIDS between haemophiliacs in MCHS and homosexual men in the 1RS 97
Years from Percent with AIDS (± se)
seroconversion Homosexual and Adults with bisexual men Haemophilia
1 0.6 (0.3) 0 (0.5) 2 3.2 (0.8) 0.6 (0.7) 3 7.2 (1.3) 3.5 (1.4) 4 13.2 (2.0) 10.4 (2.5) 5 19.9 (2.9) 15.4 (3.2)
6 32.6 (4.5) 23.6 (4.8) 7 39.8 (6.8) 27.3 (7.4)
61 2.5.3 Homosexual men and injecting drug users
Data on 350 HIV positive subjects enrolled in the Italian Seroconverter Study were analysed Of these 261 were injecting drug users (54 women) and 89 homosexual men, all with a previous antibody negative test result. The date of seroconversion was estimated as the mid-point between the last negative and first positive test dates. Analysis presented were censored at the end of 1988 with those who were AIDS-free at the time of the last medical examination more than 6 months earlier, still considered AIDS-free at the end of 1988 unless notified to the National AIDS Register.
The mean age of injecting drug users was 26 years (range = 15-48) and 35 years for homosexual men (range =16- 59). Progression rates 4 years after seroconversion did not differ significantly between the two groups: 13.8% for the injecting drug users (95% CI= 5.6- 22%) and 16.2% for homosexual men (95% CI= 5.4- 27%).
Reporting on a cohort from Turin within the ISS, Siniccoet al evaluated the risk of AIDS following the diagnosis of primary HIV infection in 74 IDUs, 33 homosexual men, 26 persons who had acquired their infection heterosexually, and one occupationally acquired infection (108 men, 26 women) All 134 subjects had a negative test within 12 months of the first positive test. Data were censored for the analysis in September 1991.
Of 23 persons presenting with acute primary HIV infection 68% had progressed to AIDS within 56 months of seroconversion (4.75 years) compared to 20% of 111 asymptomatic persons who had progressed to AIDS within 66 months (5.5. years) of seroconversion. The overall progression rate to AIDS was estimated to be 26% within 5.5 years of seroconversion.
2.5.4 Haemophiliacs and transfusion recipients
In another study from Sweden, investigators pooled data from infected haemophiliacs and transfusion recipients Beginning in 1984 all known haemophiliacs in Sweden (n=350) were screened for HIV and 100 were found to be infected. As data on 2 persons were incomplete, neither of whom had developed AIDS, 98 persons were
62 included in the analysis. The authors added that there was no evidence that any haemophiliacs in Sweden had died of HIV related illness before the entire group was screened.
In 1986 an attempt was made to find all HIV infected blood transfusion recipients in Stockholm. The donations of 35 infected donors were traced to 349 recipients of whom 180 were still alive. In total 50 living infected recipients were found but the records of only 48 were included in the analysis.
The estimated years of seroconversion were from 1979 to 1986 for the 98 haemophiliacs and 1982 to 1985 for the 48 transfusion recipients. Transfusion recipients were found to progress significantly faster than haemophiliacs (Table 2.10).
2.5.5 Transfusion recipients, haemophiliacs, and homosexual men
The Transfusion Safety Study was set up in the USA as a multicentre study to investigate transfusion-transmitted infections, and in particular HIV. HIV disease progression was studied in homosexual blood donors, infected transfusion recipients and persons infected through receiving contaminated blood clotting factors. Clinical observations began in August 1985 and ended in May 1993
Donors were traced in New York, South Florida, San Francisco and Los Angeles and sera collected from late 1984 to early 1985 were stored and then tested for HIV antibodies early in 1985. A total of 291 donors with infected donations were identified. Of these, 8 had died, 35 were lost to follow up, 9 lived too far to be included in the investigations, and 91 persons declined participating in the study. In total, 148 persons were enrolled into the study between November 1985 and March 1988, of whom 112 reported homosexual contact since 1979 and were thus included in the analysis. The median date of infection for these donors was imputed as July 1983, later than that for homosexual men in the SFCC of mid- 1981. This appeared to be reasonable as the blood donors reported fewer sexual partners than the men in the SFCC, and only 3 had been diagnosed with AIDS at study entry, indicating that infection was relatively recently acquired.
63 Table 2.10 Progression rates to AIDS in haemophiliacs and transfusion recipients in Sweden
Time from Percent with AIDS (s.e.) seroconversion (yrs) Haemophiliacs Transfusion recipients
3 0 16(6)
4 1(1) 23 (8) 5 5 (2 ) 29 (22) 6 9 (4 )
64 The potential recipients of all implicated donations were then contacted through their physicians and only after the relevant physician’s consent. Of 445 recipients of potentially contaminated blood, 279 (63%) had died and 11 declined participation. Of 136 persons enrolled, 14 were found to be antibody negative, 4 were neonates, and 6 persons reported other risk factors for acquiring HIV infection, leaving 112 persons included in the analysis. The date of infection was taken as the date of transfusion and this spanned the period July 1984 and March 1985. Data on 1237 haemophiliacs enrolled from 6 centres in the US beginning August 1985 were also available to the investigators. Of these, 684 were found to be HIV positive at study entry and 6 subsequently seroconverted. Data on 519 persons positive at study entry and with Haemophilia A who had been treated with Factor VIII concentrate since 1979 were included in the analysis. The median date of infection for these persons was imputed as July 1982 as the majority of haemophiliacs appear to have been infected 1981 to 1983.
The median age at infection was highest for the transfusion recipients- 61 years (range = 2- 88 years), with median ages of 29 years (range= 17- 62) and 22 years (range= 1- 68) for homosexual men and haemophiliacs respectively.
By May 1993 AIDS was diagnosed in 37 (33%), 33 (29%), and 165 (32%) of blood recipients, homosexual men, and haemophiliacs respectively. By 8 years following seroconversion, the cumulative risk of developing AIDS was 51%, 36% and 24% for the three groups respectively.
The authors, recognising that these analyses were entirely dependent on the estimated date of infection for the donor and haemophilia groups, carried out other analyses independent of the estimated duration of infection adjusting instead for CD4 cell values (as time-dependent covariate). When age at infection was included as a continuous covariate in the Cox regression analysis, and CD4 cell count and AZT use as time- dependent covariates, the risk of developing AIDS was similar between the transfusion recipients and haemophiliacs but higher among homosexual men. Omitting KS as an AIDS defining disease removed any significant differences.
A one year increase in age was associated with a 3% increase in hazard for the subsequent risk of AIDS.
65 2.5.6 Men and women
Data collected from 16 centres participating in the Italian Seroconversion Study eensored in December 1992 were analysed The authors restrieted the analysis to persons with at least one available CD4 cell count who seroconverted after 1985 (persons who seroconverted earlier are likely to have been infrequently sampled for CD4 cell counts and this may lead to an under-estimation of progression rates to a CD4 count of 200/ml). Of 854 persons, 321 were women (195 IDUs and 126 heterosexually exposed), and 533 were men (476 IDUs and 57 heterosexually exposed). Median ages were similar between the sexes: 24 years for women (range = 14- 51) and 25 years for men (range =16-59).
No evidence of a difference was found in progression rates with estimates at 7 years following seroeonversion being: 25% for women (95% CI= 13.8- 35.5%), and 23% for men (95% CI= 15.6- 30.4%).
Concluding remarks to Chapter
Age appears to be a good predictor of HIV disease progression, with younger adults progressing more slowly than those infected at older ages (over 30- 35 years). This finding has been reported from almost all cohorts with a wide enough age range, and in particular, cohorts of haemophiliacs and transfusion recipients. Most cohorts of homosexual men and injecting drug users have probably been unable to make this observation because their age profiles have tended to be too homogeneous.
Although progression rates appear to be slower for haemophiliacs and IDUs than for homosexual men, differenees seem to be accounted for when age, the diagnosis of Kaposi’s sarcoma, the excess mortality in the first two groups, and differenees in loss to follow up rate have been taken into consideration.
Study methodologies vary considerably. The inclusion criteria, how the cohort is assembled, and who is excluded from analysis, and in particular those who die soon after seroconverting all differ between cohorts. Methods of follow up also vary and in particular so do assumptions made about persons lost to follow up. As most of these
66 methodological details are not made explicit by the investigators in publications, comparison of progression rates between studies and over time may not, therefore, be appropriate from publications.
The AIDS incubation period is long and variable, and while we have better information on its left-hand tail and median, we do not yet have sufficient information on the right- hand tail, nor of bow much the whole distribution will change either as the management of HIV infected persons through the administration of drug regimens evolves, or through the evolution of the virus itself Further, very few studies have information on persons who have acquired their infection heterosexually.
These questions can only be addressed through the long term follow up of persons, with reliably estimated times of seroconversion, who have been infected at different calendar times and in different ways.
67 CHAPTER III
METHODS
Funding was obtained from the Medical Research Council (MRC) to set up a register in the UK of HIV-infected individuals in whom the date of seroconversion is known with reasonable precision (seroconverters). The primary purpose of the Register is to enable the distribution of times from seroconversion to AIDS and to death to be estimated, to assess whether there have been any changes in these periods over calendar time, and to examine other factors likely to influence these periods and affect any changes.
The Register is intended to be complete and ongoing to provide constant and regular monitoring of these periods. It is co-ordinated by the MRC HIV Clinical Trials Centre in collaboration with the PHLS AIDS Centre at CDSC, the MRC HIV Epidemiology Co ordinating Centre at UCLMS, and the Department of Primary Care and Population Sciences at the Royal Free Hospital School of Medicine. It is intended that it will be a shared national resource, for use by those designing studies to help improve understanding of HIV pathogenesis and of immunological and virological markers of infection and their relation to prognosis.
An Executive Committee was set up comprising the grant holders (Professor Anne Johnson, Professor Andrew Phillips, Professor Janet Darbyshire and Dr Barry Evans), Dr Abdel Babiker and the Project Co-ordinator. Their remit is to ensure the setting up and management of the register, to ensure that confidentiality of patients is maintained throughout, to supervise the Project Co-ordinator, and to report regularly to the Steering Committee and the MRC on progress. A Steering Committee was set up with the responsibility of reviewing overall progress, developing proposals for additional studies, ensuring that the organisational structure of the study remains appropriate and deciding on the general policy for publicity of the study and for the exchange of information with other interested parties. The names of all committee members appear in Appendix I.
68 I started working as Project Co-ordinator for the Register in October 1994. I have the responsibility for identifying eligible subjects, obtaining demographic and clinical information and setting up mechanisms for continually identifying new subjects. In order to obtain reliable and regular estimates of the AIDS incubation period distribution, regular follow up procedures were to be set up whilst ensuring that loss to follow up rate is kept to a minimum. The following summarises the work involved in setting up the Register and following up registered subjects.
3.1 Definition of a seroconverter
For the purposes of the Register, a seroconverter is defined as an HIV seropositive person aged 16 years or over on whom an HIV antibody test was performed and found to be negative no more than three years prior to the first positive antibody test. In the absence of a negative result, persons identified during the acute infection stage, namely when infection has occurred but before full seroconversion, are also included in the study. Laboratory evidence for this acute infection is required, the details of which were reached by consensus among expert virologists at the request of the Steering Committee (Appendix III). The Register is not actively recruiting haemophiliac patients as they are followed up through the UK Haemophilia Centres Directors' Organisation.
A maximum period of three years between the negative and positive tests was chosen to allow the choice of using a narrower period as appropriate.
3.2 Identifying study subjects
3.2.1 Clinical centres
All clinicians from Genito-Urinary Medicine, Infectious Disease and other specialities from centres in the UK taking part in the MRC clinical trials in HIV infection were invited by individual mailings to participate in this study. They were asked to help in identifying eligible patients. Additionally, further clinicians were also identified whose clinical centre had reported a cumulative total of 10 or more AIDS cases to the two national AIDS
69 surveillance centres (CDSC and SCIEH) but who were not taking part in the clinical trials. These were also invited to participate. Reporting of AIDS cases to CDSC and SCIEH started in 1982 and is estimated to be 87% complete
I visited a number of centres and met with clinicians and research nurses, laboratory and computing staff and data managers to discuss with them the objectives of the study and explore how best to identify eligible subjects at their clinics and the reporting mechanism most suitable to their set-up.
3.2.2 Laboratories
Laboratory directors and medical microbiologists reporting newly diagnosed HIV infections to CDSC and SCIEH were also contacted by letter and telephone and asked to help identify eligible subjects. They are participating in the same voluntary reporting scheme as for AIDS case reports and included National Health Service, Public Health and private laboratories, as well as regional Blood Transfusion Centres. Laboratory reporting to CDSC and SCIEH of newly identified HIV infection began in the latter part of 1984 and is estimated to cover around 80% of known infections
In October 1986 the PHLS Collaborative Laboratory Study on HIV infection was set up to gather information on all HIV testing and is therefore able to identify repeat tests on the same individual and hence incident infections. It comprised 9 laboratories and expanded to include 18 laboratories by the end of 1989 Colleagues working on the PHLS Collaborative Study at CDSC identified eligible seroconverters and provided information (patient identifiers, as well as negative and positive test dates) on them.
In Scotland, a surveillance mechanism was set up in November 1988 through which all laboratories performing HIV testing report centrally to SCIEH; the ‘Scottish Denominator Study’ It facilitates the search for any previous negative tests for a newly diagnosed HrV-infected individual. Colleagues at SCIEH provided a list of all persons known to them to be seroconverters through that system together with information on: the clinic where the latest CD4 count was requested, the name of the clinician requesting it, or the name of the clinician who had reported the patient as having been diagnosed with AIDS or
70 died. The appropriate clinician was then contacted enclosing a list of patients identified in this manner and asked to register them by providing relevant information on the registration forms. They were also asked for details of other seroconverters known to them whose seroconversion may have pre-dated November 1988 or whose match had not been made because of inconsistencies in the identifiers used in the matching process by SCIEH.
3.3 Detection of duplicate reports
Clinical reporting of AIDS cases and laboratory reporting of HIV infection at both CDSC and SCIEH run independently of one another and the use of a soundex code (a code derived fi*om the patient's surname comprising of a letter followed by three digits) as well as the date of birth allow the link to be made between an HIV and an AIDS case report on the same individual, and the elimination of possible duplicates In this way it is also possible to regularly cheek for duplicate reports of the same seroconverter on the Register.
3.4 Retrospective ascertainment
The current CDSC HIV laboratory reporting form, which has been in use since March 1993, contains a specific question concerning whether a negative test had previously been performed. For all reports where this item of information was provided, the reporting laboratory was contacted and asked to provide the name and address of the clinician looking after the patient, or, if this was not known, the name and address of the clinician requesting the positive HIV test. The clinician was then asked to register the patient by completing a registration form and also to identify and register all other seroconverters fi*om their centre including those known to be dead, transferred to other centres, or have become lost to follow-up.
Prior to March 1993, there was no specific question regarding a previous negative test on the CDSC form. All laboratories who had reported new HIV infections first diagnosed since January 1990 (as an arbitrary cut-off date) were also asked to check for any previous negative tests for persons with a positive result between January 1990 and March 1993 who were not already known to have had a previous negative test.
71 In order to identify subjeets eligible for the Register information was sought jfrom the local laboratory performing the HIV testing wherever possible. The rationale being that this method was least likely to be subject to bias, as it is independent of the patient's vital status and attendance at a clinic (see section 3.6.2.1). As this was not always possible, clinic records were also used; these may give biased estimates towards survivors if they include only current attenders, but seroconverters with a negative test performed at another laboratory/ clinical centre can also be identified by this method. A number of centres requested help with this task and this entailed going through patient ease notes myself in order to identify eligible subjects and to extract the relevant items of information.
A number of laboratories and clinical centres, particularly those with large case loads of HIV infected persons, include information on previous negative tests on their patient databases. Computerised records from the clinic were used whenever available. In order to identify eligible subjects who had died or transferred to other centres before the computerisation of records, the search was supplemented by case note review where feasible.
3.5 Prospective ascertainment
Once the mechanism for identifying and reporting the retrospectively ascertained cases was set in place the identification of new seroconverters at the clinical centres became easier. HIV and AIDS reports to CDSC and SCIEH are regularly reviewed for seroconverters who may not have been reported to the Register. Computer files are imported into the Register database at the Clinical Trials Centre and matched against existing records. Any persons not already on the Register database are thus identified. The reporting laboratory is then contacted and asked for the appropriate clinician. Clinicians are then contacted with a list of potential seroconverters and requested to register them.
3.6 Information collected
Centres are asked to complete a registration form on each eligible seroconverter. A number of the larger clinics, particularly in London, preferred to give this information as a
72 computer file directly from the clinic’s database. Any requested information not available on tbeir database was supplemented on paper at a later stage.
The following information is requested:
Patient identifiers Soundex code, initials, date of birth, and clinic/ hospital number
Epidemiological sex, ethnic group, probable route of virus transmission (i.e. exposure category: sex between men, sex between men and women, injecting drug use, other category), and likely country of infection
Clinical presence of a seroconversion illness (with date and main symptoms); the dates of last negative and first HIV positive antibody tests and where these were performed; absolute values for the two latest CD4 counts; plasma viral load measurements and dates; information on anti-retroviral drugs; drugs used as prophylaxis for opportunistic infections (names and date of start of each and when last on); whether AIDS has been diagnosed (with date(s) and disease(s)), vital status and, if dead, date of death
Information about the nature and completeness of any existing local databases was sought at an early stage of the study from the major clinical centres through discussions with tbeir computer staff or data managers. On the basis of this, the study registration and follow up forms were designed so as to make them as compatible as possible with the information collected routinely by the clinical centre (Appendix IV). This was of particular relevance to information on drugs prescribed. At registration and follow up only information on the date of start of each drug prescribed and the date last prescribed is requested.
Strict confidentiality is maintained. No names or addresses of seroconverters are requested. The database for the study is held within a secure building with separate coded access and data are completely inaccessible except through the use of passwords known to
73 key individuals only. Data released to other researchers, which would require the approval of the Steering Committee, contains no information that could lead to the identification of seroconverters.
3.6.1 Follow up information
3.6.1.1 Annual follow up
Follow up data is collected on an annual basis through the appropriate clinical centre. Information is sought on:
• the value and dates of the two latest absolute CD4 counts • plasma viral load measurements (requested since July 1997) • whether the patient is currently taking the same drug(s) previously prescribed • any additional drug(s) prescribed/ stopped in the interval since last follow up (temporary changes in medication in the intervening period are not recorded) • whether a first or subsequent AIDS-defining disease(s) has been diagnosed • whether the patient has died • the date they were last assessed clinically
If the patient was last known to he alive at a later date but their AIDS status was not known at that point, this date is also recorded. If the patient has died, the date of death is requested and details on cause(s) and any terminal illness(es) if available. Any information on the patient transferring to another clinical centre is also recorded so that subsequent follow up is done through that centre.
3.6.1.2 MRC clinical trials databases
The first MRC clinical trial in HIV infection, Concorde, was initiated in 1988 and was jointly co-ordinated with the French INSERM Clinical Trials Centre A number of other clinical trials in adults and children followed over the ensuing years to assess the possible benefits of various combinations of drug regimens over different combinations
109-113 number of seroconverters on the Register are also participating in one or more of
74 the MRC trials. This facilitated searching the trials databases for more up-to-date information as well as any information missing jfrom the Register database. The following adult trial databases were searched: Concorde, Alpha, Opal, Delta and Quattro.
3.6.1.3 The CDSC and SCIEH databases
On an annual basis AIDS and HIV infection databases held at CDSC and SCIEH were also searched with the help of staff at both institutions matching on the soundex code and date of birth of seroconverters. Other information was then used to verify the authenticity of a match, e.g. the patient’s initials, sex, clinic number and HIV test dates.
3.6.1.3.1 AIDS cases and deaths
The AIDS databases both at CDSC and SCIEH were used to identify AIDS cases and deaths occurring in patients who have stopped attending their clinic. The Register database was then updated with this information including information on the latest care- providing clinical centre so that subsequent follow up could be done through it.
3.6.1.3.2 CD4 count surveillance
A national laboratory-based surveillance scheme set up in 1991 in Scotland and in 1993 in England and Wales to monitor trends in immunosuppression associated with HIV infection Reporting to CDSC and SCIEH is directly from the laboratory performing the CD4 test and includes the following information: soundex code, clinic/ hospital number, date of birth, first name initial, sex, source hospital, the CD4 absolute cell count and date.
Of 65 such laboratories in England and Wales, 54 (83%) have participated in the scheme and provided the relevant information (R. Gilbert, CDSC- personal communication). In Scotland, all 5 laboratories undertaking CD4 counting work are participating.
Seroconverters lost to follow up were crossed checked against these databases using soundex code and date of birth matching procedures. For all positive matches, the date of the latest CD4 count was used as the date the patient was last known to be alive. This
75 information could then be used for the analysis of seroconversion to death but not to AIDS as no assumptions were made about the patient being AIDS-free or having an AIDS disease diagnosed at the date of that CD4 count. Further annual follow up was then done through the clinical centre indicated by the CD4 surveillance information.
3.6.1.3.3 The survey of prevalent clinically diagnosed HIV infection
Health districts in England, Wales and Northern Ireland were surveyed in 1996 to collect information about persons with diagnosed HIV infection who received care through their services in 1995. The service provided demographic and epidemiological information about the prevalent caseload by area of residence, and the extent to which patients with diagnosed HIV infection travelled to obtain care related to it
The following data were requested for every patient; soundex code, date of birth, sex, date last seen in 1995, health district or postcode of residence, exposure category, clinical stage of HIV disease when last seen, and ethnic group.
13,362 individuals were reported as resident in England, Wales and Northern Ireland and receiving HIV-related services in 1995. Seroconverters lost to follow up were crossed checked against these, using soundex code and date of birth matching procedures. For all positive matches, the date last seen by the health district was used as the date the patient was last known to be alive. As information on clinical stage of disease from this data source at times conflicted with information from the clinic it was ignored and no assumptions were made about an AIDS diagnosis for seroconverters with matching records.
3.6.1.4 ONS data
In England and Wales all deaths are registered locally and returns are made nationally to the Office for National Statistics (ONS) (formerly known as the Office of Population Censuses and Surveys (OPCS)). Computerisation of these records began on 1 January 1993. Computer files were available on persons who died on or after that date with the following data items: full name, date of birth, sex, date of death, and local authority
76 where death took place. After agreement fi*om ONS, a computer programmer from the Trials Centre wrote a programme which read the ONS data, with surname included. The ensuing output file, which could then be taken back to the Clinical Trials Centre, contained the soundex code in lieu of the surname.
It is estimated that the number of deaths in England and Wales approximates 600,000 every year and that approximately 540,000 (90%) of these are aged over 55 years at the time of death. In order to facilitate and speed the matching process only data on persons aged 55 years or less were extracted fi*om the ONS files. As 95% of AIDS cases and 97% of HIV infection reports in the UK are in persons aged less than 55 years, it was felt unlikely that many deaths would be missed through this exclusion. And, in any case, the death certification of persons aged 55 years or over at the time last seen who become lost to follow up may be searched for manually at a later date.
It is the intention to set up similar procedures with the General Registrar Office (GRO) in Scotland. Negotiations are, however, at an early stage.
3.6.2 Supplementary information
3.6.2.1 How seroconverters are ascertained
Information on how I identified seroconverters was recorded, whether through the laboratory performing the HIV test or through the clinic requesting the test. This information was not always readily identifiable, however. Seroconverters who were reported to the Register in direct response to myself having initially identified them fi*om laboratory reports to CDSC and SCIEH, were recorded as identified through the laboratory. In retrospect, it is possible that a number of these may, in fact, have been originally identified through the clinical centre who furnished the laboratory with this fact at the time of making the HIV test request (which tested antibody positive). The laboratory simply relayed this information to CDSC. It proved almost impossible, however, to disentangle this information.
77 Seroconverters reported to the Register from the clinical centre without prompting were coded as identified from the clinic. This included persons who had apparently tested negative elsewhere other than the centre now providing the clinical care. Where this information had also been reported to CDSC by the laboratory, they were recorded as ‘both’ (clinic and laboratory). In Scotland all seroconverters save those from the Edinburgh City Hospital Clinic, were identified through the Scottish Denominator Study and coded to ‘laboratory’.
This information was documented in order to evaluate whether information on seroconverters identified by the clinic were likely to be biased towards persons presenting with a severe illness, or to those progressing more rapidly. Such a bias would lead to faster progression rates. Information on seroconverters identified through the laboratory, we reasoned, was less likely to be subject to such recall bias if ascertainment simply depended on the fact that a previously negative antibody test had been performed on the same attender.
3.6.2.2 Documentation of a previous antibody negative test
The reliability of information on a previous antibody negative test to HIV varied across centres and over time. At the time when an attender requests an HIV test at a clinical centre, risk factor information is generally solicited at a counselling session and often includes a question on whether the attender had previously had such a test. The attender may have visited the clinic previously on occasions for other sexually transmitted diseases and information on a negative HIV antibody test performed at the clinic would generally be noted in clinic records. It was important to establish for each person reported to the Register confirmation of a negative test having been done as well as the timing of it.
Information was requested on the date (day and month included) and place of testing. Wherever information was provided that the negative test had been done at the clinic now reporting a seroconverter, such a seroconverter was flagged as ‘documented’. This rule applied given the reporting clinic was able to provide at minimum the month of the negative test date. Where only the year of the test was provided, and/or a different clinic given as the place of the negative test, the seroconverter was flagged as ‘undocumented’
78 (unless correspondence from that clinic substantiated this fact), or ‘pending documentation’ if no clinic was given as the site of the negative test. Negative tests carried out overseas were recorded as such, unless documentary evidence was supplied, which happened rarely. Supplementary information was then sought to check the validity of the negative test date for those remaining undocumented. Information was sought from other clinics and through the laboratory reporting system at CDSC. Information on negative test dates obtained from the Denominator studies at CDSC and SCIEH were always taken as ‘documented’. In this manner a number of negative test dates previously only reported by the patient could be authenticated, or corrected. This process is continual.
3.6.2.3 Completeness of identification of eligible attenders at clinical centres
The aim of the Register at the outset was that it should be a complete record of all known seroconverters in the UK, so as to minimise ascertainment bias. For a number of reasons, however, this was not achievable. When the Register was being set up in October 1994 information was rarely available in many clinics for non-current attenders, the majority of whom were likely to have died. It was important to document the completeness of seroconverter registration at each clinic so that it was known whether: a) all possible seroconverters have been identified and reported regardless of their current vital status (i.e. the cohort includes current attenders, those who have died, and those who have become lost to follow up). b) all possible seroconverters have been identified who were attending the clinic from a given point in time, say 1st January 1990, so someone who had died or stopped attending the clinic before that date would not be included. e) seroconverters were haphazardly recalled to mind by staff at the clinics and/or laboratories.
From these data a ‘date of entry’ into the cohort could be ascribed for each seroconverter reported to the Register (see section 4.5). A listing of the completeness of ascertainment, and hence, date of entry, is given for each centre in Appendix V.
79 3.7 Blood store
Clinicians were also asked, with informed and written consent from the patient, to take 10 ml blood samples annually for central serum storage. Local Ethics committee approval is necessary and help with filling in the local Ethics Research committee forms was offered. All clinical participating centres obtained local ethics approval.
A further 30 ml sample was also requested until mid-1995 from patients who had seroconverted more recently (defined as having the first antibody test in the preceding 6 months and with a negative test within one year of the first positive test). This was sent to one of two MRC repositories in Cambridge and Edinburgh as part of the MRC Molecular Epidemiology Programme (MEP). As information sought on patients for the Register is similar to that sought by the repositories, and in order to save duplicate form filling at the clinical centre, a simple form was designed to accompany specimens sent to the repositories. This alerted the virologist at the repository that the sample was taken from a patient participating in the Register (and therefore, by previous agreement, half of the quantity of plasma, sera and cells could be stored in the Register’s central clinical storage) and information on these seroconverters can then be relayed back to them from the Register’s database.
80 CHAPTER IV
METHODOLOGICAL ISSUES
There are a number of possible sources of bias in any cohort study. It is important to identify all potential sources so that, wherever possible, they can be minimised when designing the study and controlled for in the resulting analyses. This chapter aims to identify all such sources, and in Chapter VI an illustration of their effect will be made, where possible, by controlling for them and then by ignoring them. This is particularly relevant when attempting to compare the published results from one cohort with those of another, or at times two results from the same cohort.
4.1 Representativeness of HIV infected persons
HIV infected persons recruited into cohorts may not be representative of HIV infected persons in general. The total number of HIV infected persons in a population is never known, but can be estimated through a number of methods using other available information. The serostatus of most HIV infected persons will become known at some stage during their infection. This may be shortly after infection, at the diagnosis of an AIDS disease, close to death or at post-mortem or at some other stage in the intervening time points. For a few it may be that HIV disease will remain undiagnosed.
Of those diagnosed as infected, most are likely to have had only one antibody test (which tested positive). Concerns about risky behaviour, illness, or the serostatus of a sex partner are likely to be the major reasons why a person seeks an HIV test. HIV infected persons who have previously tested negative (repeat testers) may, therefore, be more likely to engage in risky behaviour, have an illness (clinical or sub-clinical) or to have a partner who is known to be HIV positive ^ The presence of an illness has been associated with more rapid progression estimates (section 4.2), and repeat testers (seroconverters), therefore, may have higher progression rates than non repeat testers (i.e. HIV infected persons who have never had an antibody negative test).
81 Persons who undergo repeat testing for HIV are not necessarily representative of all HIV infected persons. For example, heterosexuals acquiring the virus through sexual contact often have not previously perceived themselves to be at risk of HIV infection and so tend not to have had previous negative tests for HIV and, therefore, are under-represented among seroconverters. From Table 4.1 it can be seen that the exposure category distribution of persons reported to the Register is more similar to that of reported AIDS cases than HIV infection reports. While sex between men and women account for 17% and sex between men account for 66% of HIV infection reports respectively, the proportions of persons reported to the Register who were infected through sex between men and women is 11% and of sex between men 77% respectively, indicating that persons who have acquired their infection heterosexually are under-represented as seroconverters. The most recent prevalence estimates for England and Wales from the Projections group^^, on the other hand, estimate that 21,200 persons (excluding haemophiliacs and transfusion recipients) were alive with HIV infection at the end of 1993 of whom 6800 (32%) were infected through sex between men and women. The proportion of diagnosed persons infected heterosexually of the total infected in that group is likely to be small compared to these proportions for homosexual men and injecting drug users.
This source of bias cannot easily be controlled for without the availability of precise estimates of the relative hazards for HIV acquisition for repeat testers versus non-repeat testers including those not diagnosed as infected. We have not, therefore, attempted to control for this potential source of bias in the analyses.
4.2 HIV-related illness as a reason to seek a test
In the course of HIV disease, illness may be primary HIV infection, and therefore indicative of recent seroconversion, or else an indication of more advanced disease.
There have been a number of studies indicating that persons presenting with an illness compatible with primary HIV infection at seroconversion progress faster than persons who do not. Pedersen et al reported three year progression to CDC Stage IV from estimated seroconversion was 78% (95% CI= 52- 100%) for persons with long-lasting primary infection (n= 27) (defined as illness with fever lasting a minimum of 14 days)
82 Table 4.1
Exposure category distribution of seroconverters, reported AIDS casesf and HIV infections^ (to the end of September 1994) and estimated prevalence^ (at the end of 1993)
Exposure category Reports to the Cumulative incidence reported to CDSC and SCIEH Estimated prevalence Register AIDS HIV at end of 1993
Sex between men 1563 (77%) 7334 (79%) 13966 (66%) 12350 (58%) Sex between men & women 216(11%) 1181 (13%) 3646 (17%) 6800 (32%) Injecting drug use 192 (10%) 563 (6%) 2645 (12%) 2050 (10%)
Other/undetermined 51 (3%) 249 (3%) 943 (4%) -
Total 2022 (100%) 9327 (100%) 21200 (100%) 21200 (100%)
I excludes persons infected through contaminated blood clotting factors, blood transfusion i;ecipients and vertically infected children so as to be compatible with reports made to the Register
oo UJ compared with 10% (95% CI= 0- 28%) for asymptomatic seroconverters (n= 40) or persons with a mild illness (n= 19). Lindbaek et al reported 95% of symptomatic seroconverters (n=19) compared with 66% of asymptomatic seroconverters (n=29) followed up after a mean of 7.2 years from seroconversion had progressed to CDC Stage IV In a study from Turin 68% of persons presenting with acute primary HIV infection (n= 23) had progressed to AIDS within 56 months of seroconversion (4.75 years) compared to 20% of asymptomatic persons (n= 111) within 66 months (5.5. years) of seroconversion Dorrucciet al reporting for the Italian Seroconversion Study, on 391 IDU seroconverters observed an increase in the risk of AIDS for those diagnosed with an acute retroviral syndrome. Compared to other seroconverters those with acute infection had a relative risk of AIDS of 5.59 (95% CI= 2.79- 11.20) after adjusting for age, sex and year of seroconversion. Dorrucci and colleagues further reported that while CD4 cell levels were similar within the first year of seroconversion, the rate of CD4 cell decline was faster in persons with acute infection than in those without.
If persons with symptomatic primary infection are more likely to seek an HIV test and become included in the study cohort, this could lead to over-estimates in observed progression rates.
When constructing the UK Register, I gathered information on the presence of a possible seroconversion illness, its date and symptoms so that this cofactor can be adjusted for in the analysis. The symptoms relating to HIV seroconversion illness (fever, lymphadenopathy, headache, ‘flu-like’ symptoms) are, however, neither sensitive nor specific enough to HIV disease. And as this information is mainly retrospectively identified in the elinic at the time of the HIV positive test, it may be rather subjective. Further, whereas the presence of illness may be noted in clinic records, its absence may indicate that the patient did not experience an illness or simply that the question was not asked or the answer not recorded.
Of 2022 seroconverters included in the analysis of the UK Register in September 1997, information was not available for 941 seroconverters on whether they had a seroconversion illness. Of 1081 persons where this information was available, 325 (30%) reported an illness. There was no evidence that the report of a seroconversion
84 illness was disproportionate by exposure category with 29.2% to 33.6% of all exposure groups, where this item of information was available, reporting an illness.
In later disease, if a person becomes ill within 3 years of the last antibody negative test date and seeks an HIV test, they will fulfil the entry criteria of being a seroconverter and thus become included in the cohort. If, on the other hand, they experienced an illness some 7-8 years after the negative test, say, they would not then be eligible for inclusion in a seroconverter cohort due to the wide window between test dates. In this manner rapid progressors may be preferentially included in a seroconverter cohort. Ideally one would wish to test all negative persons frequently, say every 3 months so that the detection of a positive test is not dependent on the attender’s reason for seeking a test. A wider window between test dates as an inclusion criteria may help overcome this source of bias (though this will be at the expense of greater uncertainty around the time of seroconversion). The width of the test interval is examined as a covariate in the analyses in Chapter VII.
4.3 Validity of the negative HIV antibody test
Information is sought on the registration form on the date of the last antibody negative test and where this had been done. As this information is crucial to reliably estimate the time of seroconversion, it was important to establish its authenticity. All persons reporting a previous negative test were entered onto the database. However, only those where the source of information on the negative test was the clinic or laboratory where the test was performed were treated as ‘documented’ evidence.
If the attender disclosed that he/she had previously had a negative HIV test at another clinic, this fact was noted together with the date of the test. Information was subsequently sought from the clinic and/or laboratory involved to verify this information and the exact date of the test.
I carried out an analysis of all persons reported to the Register with a maximum window between test dates of 3 years using the documentation of the antibody negative test as a CO factor in a Cox model. Results are presented in Chapter VI.
85 All analysis giving results of progression estimates and associations of the cofactors examined use data only from documented seroconverters, however.
4.4 Estimating the time of seroconversion
No matter how well a cohort is followed up and how frequent follow up visits take place, seroconversion is rarely observed. Seroconversion will have occurred at some time between the last visit that the cohort member was antibody negative and the first visit that he/she was antibody positive. Seroconversion is therefore interval censored, and the wider the interval the larger the uncertainty around the actual date of seroconversion.
The most common and simplest method for estimating the date of seroconversion is by simply taking the mid-point between the two test dates. The maximum possible error in estimating the date of seroconversion will, therefore, be equal to half the width of the interval between the negative and positive test dates. In analysing data available from the International Registry of Seroconverters, Biggaret al found that persons with an interval of greater than 24 months had significantly different AIDS incubation curves from persons with narrower intervals In the first 5 years following seroconversion, the rate of progression in those with intervals > 24 months were lower than in other groups, but after 5 years progression rates were much faster.
Other methods of estimating the time of seroconversion are generally used by investigators when the interval between negative (real or imputed) and positive tests exceeds 24 months. These methods give rise to seroconversion estimates for the cohort as a whole rather than for individuals within it. Darby et al first estimated the cumulative distribution function of seroconversion times, then the distribution of séroconversions in calendar time conditional on the individual having been tested The estimated proportion seropositive is zero prior to December 1979 (the earliest recorded positive test date in their data), and unity from October 1986 (the last recorded negative date for a person who subsequently seroconverts). The probabilities of seroconverting per unit time within each of 3 defined calendar time periods from 1 January 1979 and 31 December 1987 were then calculated.
86 Byers et al back estimated the date of seroconversion for men in the SFCC who were seropositive on entry into the cohort using a probability estimate based on men with 24 months or less between dates of negative and positive serum specimens
HIV infected persons are eligible for inclusion in the Register if the window between test dates does not exceed 3 years. Using the mid-point method for estimating the dates of seroconversion, then, will lead to a maximum possible error of 1.5 years. Errors in estimating the date of seroconversion using this interval will be particularly relevant to progression estimates within 2 years of seroconversion. This can be overcome by using data from persons with narrower intervals to estimate these progression rates. Problems arise, however, in the choice of too narrow an interval as this may contain a peculiar group of individuals frequently undergoing repeat HIV testing and may be at higher risk of progression (discussed in section 4.1).
The restricting of seroconversion intervals is explored in analysis in Chapter VII. The impact on progression rates particularly in the first two years following seroconversion is examined.
4.5 Left truncation of survival times from seroconversion
Serious bias may arise if the likelihood of a person being included in a cohort is dependent on how rapidly that individual develops AIDS and/or dies. For example, in some centres seroconverters who have progressed rapidly and died may be less likely to be registered than those who are currently attending. This problem arises when a clinic/laboratory does not have a complete record of all patients ever seen. In such centres data on, for example, a person who seroconverted in 1987, and enters the cohort in 1995, cannot be used to derive information on the first 8 years of the incubation period since, if the patient had died within 8 years (i.e. before 1995), they may well have not been included in the cohort. Such a patient can thus only be entered in the risk set for the analysis in 1995 at 8 years. This is referred to as “late entry” or “left truncation” If such a patient had attended at a clinic or was identified through a laboratory with complete records for every patient seen/tested since, for example 1990, they enter the risk set at 3 years from seroconversion (1990 - 1987 = 3 years) since survival or death after this date will not influence whether or not they are included in the cohort.
87 Therefore, for individuals from centres whose ascertainment is believed to be complete from laboratory and/or clinic records (that is, all patients alive, dead, transferred and lost to follow up are included), follow up commences from each individual’s seroconversion time. However, for centres where ascertainment is judged to be complete only from a given date, h, seroconverters are treated as Tate entry’ subjects and each enters the risk set at Qi - estimated date of seroconversion). If ascertainment from a centre is not complete from any point in time, or is uncertain, follow up for all seroconverters from that centre commences from the date of registration. Thus, for these subjects their earlier contribution to estimates of the IPD may be unnecessarily excluded. For subjects with a negative test at another centre or who are already known to be HIV infected when transferring, follow up (i.e. entry into the risk set) commences from the date they first attended the clinic with complete records on all patients. If this is uncertain, or if neither clinic have complete records the follow up of such subjects commences from date of registration. Patients transferring into a clinic without complete records from another, also with incomplete records but from an earlier time, enter the risk set from the time of transferring, and not the earlier time. For example, a person seroconverting in January 1986 who is attending clinic X where records are complete only from January 1990, say, transfers to clinic Y where records are complete from January 1993. Had they been reported from clinic X they would enter the risk set at 4 years (1990-1986= 4 years) following seroconversion. If not reported by clinic X, and clinic Y reports them, they then enter the risk set at 7 years (1993-1986= 7 years) following seroconversion.
To give examples of this potential bias, consider the following scenarios:
• In clinic A, it is possible to identify all HIV infected persons who have ever attended the clinic regardless of whether they have since died, transferred to other centres or are currently alive and attending at this clinic. • In clinic B, records became computerised for patients attending from January 1990 onwards, and are complete for all attenders since then. Therefore, eligible subjects who had died or transferred to other centres prior to January 1990 were not entered onto the system and cannot, therefore, be identified. • In clinic C, complete records for attenders are available from January 1988, and so patients who had died or transferred to other centres prior to that date are not included. For patients in clinic A follow up can be backdated to the date of their individual HIV positive tests and their date of entering the risk set is, therefore, the estimated date of seroconversion. For patients in clinic B follow up can only start from January 1990 since information is incomplete for the period prior to this date and their follow up commences from that point so that they enter the risk set at the number of years after seroconversion they were in 1990. Similarly, patients registered from clinic C enter the risk set at the number of years after seroconversion they were in 1988 and their follow up begins from that point.
To illustrate how the problem of late entry is dealt with, let us suppose that 100 eligible seroconverters are identified from each of the three centres A, B and C and their estimated dates of HIV seroconversion are as illustrated in Table 4.2. The resulting lifetable for analysis would resemble Table 4.3.
An illustration of the importance of correctly allowing for late entry in analyses is the extreme case of estimating the AIDS incubation period from persons diagnosed with AIDS (as was the case for early estimates of the IPD using data from transfusion - associated AIDS cases). If persons are ascertained through being diagnosed with AIDS they then enter the risk set for an analysis of the incubation period at the date of the AIDS diagnosis. In this manner they would effectively be excluded from analysis as they contribute no information to the time from seroconversion to AIDS. The effect of late entry is explored in Chapter VI (section 6.6). Analyses are performed allowing for late entry and results are then compared to the results from analysis where late entry is ignored.
4.6 Right censoring
In a cohort where the occurrence of a particular event is the outcome measure, those persons who remain event-free throughout the period of observation are censored at the date that follow up for the study terminates. Persons who have died or become lost to follow up during the period of observation are also censored (assuming death is not the outcome of interest) and typically this is at the date they died or became lost to follow up. This may not be always appropriate, however, and in this section I discuss the reasons for this and what alternative strategies may be employed.
89 Table 4.2 Late entry: number of patients recruited from each of 3 hypothetical clinics
Estimated year of Clinic seroconversion A B c 1984 20* 0 25 1985 10 25 25 1986 0 0 0 1987 20 0 25 1988 0 0 0* 1989 25 25 0 1990 0 0* 0 1991 0 0 0 1992 20 25 25 1993 0 0 0 1994 5 25 0 Total recruited 100 100 100
* Time h from which records are believed to be complete. Patient numbers above these are treated as Tate entry’ subjects in the analysis. Table 4.3 Late entry: life table of retrospectively identified seroconverters from hypothetical clinics A, B, and C (see Table 4.2)
Time from number of number Cumulative event- seroconversion patients in risk of events free survival (years) set probability 0 175* 10 0.94 1 165 + 25t + 2 5î 10 0.90 2 205 15 0.83 3 190 + 25$ 6 0.81 4 209 + 25$ 10 0.78 5 224 + 25$ 12 0.74 6 237 7 0.72 7 230 10 0.69 8 220 14 0.64 9 206 13 0.60
* 100 patients from clinic A, 50 from clinic B, and 25 from clinic C t Late entry subjects from clinic B J Late entry subjects from clinic C
90 4.6.1 Persons lost to follow up
An important consideration in survival analysis is the proportion of persons who are censored because they were last seen before the cut-off date for the analysis and were event-free at the time they were last seen. To illustrate the effect of these persons on analysis, consider the following scenario of patients A-J in a cohort study. Their contribution to events and person-times at risk are illustrated in figure 4.1 : Patients B, C, and H experience the event of interest during the period of observation. Patients E and I are under observation throughout the whole period and remain event- free, and are therefore censored at the cut-off date for the analysis. Patients A, D, F, G and J are last seen event-free at some time before the analysis cut-off date and are typically censored at the date they were last seen/evaluated for the study.
The reason why the last 5 individuals were not seen for some time before analysis cut off date may simply be that this was their latest appointment and no further appointments were scheduled until after the end of the period of observation. This may be particularly true of patient G, for example because their last clinic attendance was just before the cut-off date for analysis. Clinic attendance, however, may be outcome dependent, i.e. whether a person attends for an appointment is dependent on how well they are and/or feel. This is often the case for HIV infected individuals where good health tends to result in less frequent clinic visits. It may be that patient A, for example, was completely free of symptoms at the last clinic attendance and no further appointments were given until sometime after the analysis cut-off date (or the patient may opt not to come for further appointments until they become ill). If this is the case then censoring is informative. Censoring those persons’ follow up on the date each was last seen may well shorten the person-years under observation for those without the event, and therefore bias results because persons experiencing the event will be over represented.
Often investigators are able to cross-check with national AIDS registers to establish whether members of the cohort who became lost to follow up attended at other centres and were subsequently diagnosed with AIDS there Following such cross linkages, a number of persons may be identified as having been diagnosed with AIDS following the time they were last seen (such as attenders F and I in Figure 4.1). If their
91 Figure 4.1
An illustration of right censoring bias
Patients
A
B
C
D
E
F
G
H
I
J
Start of period Analysis Cross-check of observation cut-off with AIDS Register
I Last clinic attendance X AIDS- while under follow-up
X AIDS- ascertained from Cross-check with CDSC and SCIEH Registers
92 records are then updated such that they are now AIDS cases, then higher progression estimates will be derived if no further information is available on persons who remained event-free after they were last seen. One way of avoiding this source of bias is to assume that persons not found to have been diagnosed with AIDS remained AIDS-ffee at the time the period of study observation ended, thus lengthening the person-years of observation for AIDS-ffee individuals.
This makes assumptions, however, about the completeness of registers used for cross linkage of persons lost to follow up, and there are a number of issues to consider in relation to this (discussed in sections 4.6.2 and 4.6.3).
A number of right censoring strategies in estimating progression rates to AIDS are explored, the results of which are presented in Chapter VI. These strategies are as follows:
• censoring persons at the analysis cut-off date, the date of a death without AIDS, or the date they were last seen AIDS-ffee, whichever is the earliest. • a number of cut-off dates are chosen and all persons who were not reported as having AIDS are uniformly censored on that date. • a cut-off date is chosen when it is reasonable to assume that persons not reported to CDSC and SCIEH by then are AJDS-free
4.6.2 Delayed reporting to the national AIDS surveillance register
It is inevitable in any system that there will be a delay between a clinician diagnosing an AIDS case and the subsequent reporting to the appropriate authorities of that case. Knowledge of the likely time lag between diagnosis and reporting is essential as this must be allowed for once matching is carried out and any assumptions made about persons who still remain Tost to follow up’.
Table 4.4 shows the distribution of the lag between diagnosis and reporting of AIDS cases reported to CDSC and SCIEH to the end of 1996. Of cases reported to CDSC and
93 SCIEH 94% are estimated to be reported within 24 months of AIDS diagnosis*. Therefore, after allowing for a 24-month reporting delay it would be reasonable to assume that persons not found on the AIDS register are still AIDS-free at the end of the period of study follow up. This may be too conservative a reporting lag and it may not be unreasonable to allow for 12 months for delayed reporting (84% are reported within 12 months of diagnosis)*.
4.6.3 The completeness of the national register
As well as delayed reporting of AIDS cases, under-reporting may also exist. The above strategies assume that all cases of AIDS are eventually reported to the national register. In the UK the national surveillance database is estimated to be 87% complete however, and so it is likely that the assumption that persons who remain unmatched are AIDS-free may under-estimate AIDS incidence in the cohort and so lead to longer estimates of the time to AIDS. Clearly the higher the proportion of reported cases to total diagnosed the greater the certainty that assumptions are correct regarding unmatched cases.
4.6.4 Deaths in persons without AIDS
In estimating the rate of progression to AIDS in a cohort, persons who die without an AIDS diagnosis are generally censored at the date of death. In this manner they are, in effect, treated as if they were lost to follow up at that date. If the risk of death without an AIDS diagnosis is low, this strategy will not have a significant impact on progression estimates. However, if the risk of AIDS in those who died is different to the risk of AIDS for persons who did not die but are at the same time from seroconversion, censoring persons who died without AIDS will give biased estimates of the IPD.
Cohorts of haemophiliacs and injecting drug users, however, have noted significant pre- AIDS mortality, with one half of the deaths in the cohort being in persons who had not
* These percentages are calculated from the cases diagnosed 1979 to 1994 (n= 1102) to avoid bias due to right truncation as persons diagnosed in 1995 and 1996 are constrained to have a maximum reporting lag of 24 and 12 months respectively.
94 Table 4.4
Reporting delay for UK AIDS cases reported to CDSC and SCIEH to the end of 1996
Year of AIDS Delay in months between diagnosis and reporting TOTAL diagnosis number of cases (proportion of total diagnoses) <3 3-5 6-12 13-24 25+
1979-1990 2278 (48%) 766 (16%) 772 (16%) 512(11%) 401 (8%) 4729 (100%) 1991-1992 1656 (57%) 444 (15%) 377(13%) 244 (8%) 202 (7%) 2923 (100%) 1993-1994 1905 (55%) 593 (17%) 462 (13%) 317(9% ) 173 (5%) 3450 (100%)
1995-1996 1456 (58%) 554 (22%) 389 (16%) 95 (4%) - 2494 (100%)
Total 7295 (54%) 2357 (17%) 2000 (15%) 1168(9%) 776 (6%) 13596 (100%)
LA been diagnosed with AIDS This will tend to result in lower progression rates to AIDS for those two groups. If the rate of pre-AIDS mortality changes significantly over time, this will result in changes in the AIDS incubation period being observed, so that if a higher proportion of persons died without AIDS in later time periods a slowing in the AIDS progression rate will be observed. It is important, therefore, to take into effect the proportions of pre-AIDS mortality in the cohort when comparing progression rates to AIDS with those from another cohort study.
4.7 AIDS definitions and diseases diagnosed
As previously discussed, AIDS is a spectrum of conditions some of which tend to occur at different levels of immune suppression than others Kaposi’s sarcoma tends to present at relatively high levels of CD4 cell counts, and for this reason time from seroconversion to the diagnosis of KS tends to be shorter than to other AIDS-defrning manifestations. This condition was first described by Professor Moriz Kaposi Kohn, a Hungarian dermatologist, in 1872 as a relatively slow-growing sarcoma on the legs or feet and occurring predominantly among elderly men aged 50-70 years of Italian or Eastern European (Ashkenazi Jew) descent. Suggestions in the 1980s of KS being caused by a transmissible infectious agent led to the discovery in 1996, of Kaposi’s sarcoma-associated herpesvirus (KSHV/HHV-8) as the probable cause for KS Its exact mode of transmission, however, has not been fully explained.
From 1985 to 1992, KS was the AIDS-defrning illness in approximately 20% of all homosexual and bisexual men with AIDS. Even more men develop the tumour at some time during the course of AIDS. This percentage has decreased as the AIDS epidemic has progressed. During the 1992-1994 period, 14% of homosexual/bisexual men with AIDS developed KS. This reduction is thought to be due to increased safer sex practices.
Of particular interest, is that all other exposure categories have much lower rates of KS. For this reason homosexual men appear to progress to AIDS faster than other exposure categories. Because the incidence of KS has decreased over time, this may have the effect of observing an apparent, but spurious, increase in the AIDS incubation period.
96 In an analysis of progression rates to AIDS, and particularly when comparing homosexual men with another exposure group, the role of a KS diagnosis is examined in Chapter VII (section 7.3). This is ideally done by ignoring KS as an AIDS defining disease, i.e. measuring time from seroconversion to the subsequent AIDS disease. In reality, information on AIDS diseases subsequent to that initially diagnosed are poorly reported. As an alternative strategy, persons with KS are censored at the date of the KS diagnosis.
97 CHAPTER V
CHARACTERISTICS OF THE UK REGISTER COHORT
In total, 142 clinical centres and 136 laboratories throughout the UK were identified and contacted. Of these, 131 (92%) and 103 (76%), respectively, agreed to participate in the Register, representing 91% of all AIDS case reports and 97% of HIV infection reports made in the UK to CDSC and SCIEH.
By the end of September 1997, a total of 2022 HIV antibody positive persons aged 16 years or more who were not known to be haemophiliacs and who had had a previous negative antibody test were reported to the Register from 96 clinical centres. However, 6 centres (1 in Scotland and 5 in London) reported 1389 (69%) of these cases (Figure 5.1).
Of 2022 persons reported, 1511 (74.7%) were identified through clinic records, 435 (21.5%) through laboratory records, and 76 (3.8%) through both the clinic and laboratory records. The HIV test interval, i.e. the time difference between the first antibody positive and the last antibody negative test dates was within one calendar month for 76 reports, and exceeded 36 months for 201 reports. Characteristics of these 2022 persons are summarised in Table 5.1.
5.1 Exposure category
Persons reported with more than one exposure category for HIV infection were assigned to a single category according to the following hierarchy: sex between men, injecting drug use, sex between men and women. For example, persons reported to be homosexual men and injecting drug users are coded to ‘sex between men’. Persons reported to have been exposed through sex between men and women and who also injected drugs were coded to ‘injecting drug use’.
The main category of exposure to HIV infection of those reported, comprising 77.3%, was through sex between men (n= 1563), followed by sex between men and women (n= 216) and injecting drug use (n= 192). For 12 persons exposed to HIV through
98 Figure 5.1 Distribution of ali subjects reported to the Register to the end of September 1997 by reporting clinical centre
Ko bier
CXH 5%
MMC RFH 13% 7 % ECHC
CXH- Charing Cross Hospital, London ECHC- City Hospital Clinic, Edinburgh Kohler- Kohler Centre, London MMC- Mortimer Market Centre, London RFH- Royal Free Hospital, London SMH- St Mary’s Hospital, London
99 Table 5.1 Characteristics of all subjects reported to the Register! October 1994 - September 1997
Number of Proportion reports of cases (n=2022) (%) Sex Male 1816 89.8
Female 206 10.2 Exposure category Sex between men* 1563 77.3 Sex between men & women 216 10.7 Injecting drug use 192 9.5 Other/undetermined 51 2.5 Seroconversion within 1 calendar month 76 3.8 Interval 1-12 months 931 46.0 13-24 months 506 25.0 25-36 months 308 15.2 >37 months 201 9.9 How ascertained Clinic 1511 74.7 Laboratory 435 21.5 Both 76 3.8 Ethnicity white 1014 50.1 black African 42 2.1 black Caribbean 17 0.8 Other 200 9.9 Not known 749 37.0 Seroconversion Yes 325 16.1 illness reported No 756 37.4 Not known 941 46.5 Possible country UK 580 28.7 of infection Other 119 5.9 Not known 1323 65.4
t includes 201 subjects with HTV test interval >36 months * includes 16 men who also reported to have injected drugs.
100 other means, 4 were health care workers exposed occupationally, 7 through a contaminated blood transfusion and one who self-injected with HIV positive blood. The exposure category of 39 persons was not reported.
Differences were observed in the distribution of exposure categories by year of seroconversion (Figure 5.2). The earliest séroconversions in 1982 to 1984 were almost all injecting drug users, while few IDUs reported to the Register seroconverted after 1986. A modest rise was observed in the proportion of persons infected through sex between men and women from 1986 to 1993. From 1985 persons exposed through sex between men continued to account for the largest proportion of seroconverters in each calendar year.
5.2 Sex
Forty-two clinical centres reported women subjects. Women comprised 10.2% of total subjects reported (n= 206). Seventy women (34%) were reported from one clinical centre in Scotland. Of the 206 women reported to the Register, 127 (61.7%) were reported to have been exposed to HIV through sex between men and women, 69 (33.5%) through injecting drug use and 7 through other means (3 health care workers with occupational exposure, 3 transfusion recipients, and one who self-injected with HIV positive blood). The exposure category of 3 women was not reported. Women injecting drug users were younger than women exposed through sex between men and women with a median age for IDUs of 23 years (range= 15-35 years) compared to a median age of 27 years (range= 17- 57 years) for those exposed through sex between men and women.
5.3 Seroconversion illness and estimated time of seroconversion
Presence or otherwise of a seroconversion illness was recorded for only 1081 persons on the Register, and of these 30% reported symptoms compatible with a possible seroconversion illness. This proportion is higher than reported by other studies and it may be artificially high because persons ‘not known’ to have had an illness may, in fact, not have been ill (and so no information relating to an illness was recorded in the notes). Because of this, illness was not included as a cofactor in progression analyses to AIDS and death.
101 Figure 5.2 Exposure category distribution by calendar year of seroconversion
300 □ sex between men □ IDU
250 □ sex between men and women □ other/ undetermined
200
150
100
50
8 The estimated year of seroconversion was calculated simply as the mid-point between the last negative and first positive antibody test dates. Estimated seroconversion ranged from 1982 to 1997 contributing a total of 8678 person-years of follow up. By September 1997, 494 seroconverters (24.4%) had been diagnosed with AIDS and 314 (15.5%) had died (Table 5.2).
5.4 Age at seroconversion
Age at seroconversion was below 30 years for 57.3% of seroconverters (n= 1158) (Table 5.3). Injecting drug users were younger than persons in other exposure categories with a median age at seroconversion of 23 years compared to 28 and 29 years for persons exposed through sex between men, and sex between men and women respectively. Because of this, women were also younger than men at the time of seroconversion with a median age of 25 years compared to 28 years for men (Table 5.4)
Two persons aged 15 years were included in the Register (the minimum age for inclusion on the Register is 16 years) because they had reached their 16th birthday by their first positive test date. The age distribution by estimated year of seroconversion reflects the younger ages of injecting drug users seroconverting pre-1985 (figure 5.3).
5.5 Ethnicity and possible country of infection
Ethnicity was poorly reported, as it was not usually recorded in the attender’s clinic record. It was generally only available for persons still attending or who have recently attended the clinic. Of 1273 persons for whom ethnicity was reported, 80% (n= 1014) were white. The country of possible infection was also poorly reported, and where reported this was generally the UK.
5.6 HIV test interval
Two-hundred and one persons had an interval greater than 36 months between the last negative and first antibody positive test dates and were, therefore, excluded from
103 Table 5.2 Total number of subjects reportedf, number of AIDS diagnoses, and number of deaths known by 30 September 1997
Estimated year of number of number number known to seroconversion subjects diagnosed with have died AIDS by 30.9.97 (from any cause) by 30.9.97 1982 7 6 5 1983 45 21 27 1984 57 28 32 1985 46 27 19 1986 71 38 37 1987 78 39 28 1988 117 51 37 1989 172 56 32 1990 215 59 36 1991 234 63 27 1992 271 51 20 1993 277 34 9 1994 208 11 3 1995 153 9 2 1996 62 1 0 1997 9 0 0
TOTAL 2022 494 (24.4%) 314 (15.5%)
t includes 201 subjects with HIV test interval >36 months
104 Table 5.3 Age distribution of all subjects reported to the Register^ by exposure category
Exposure Age group at estimated seroconversion (years) category <20 20-24 25-29 30-34 35-39 40-44 45-49 50+
Sex between men 53 328 487 329 187 98 38 43 Sex between men and women 13 46 58 44 22 13 8 12 Injecting drug use 31 76 47 26 8 4 0 0 Other/undetermined 2 7 10 8 15 7 0 2
TOTAL (100%) 99 457 602 407 232 122 46 57 (4.9%) (22.6%) (29.8%) (20.1%) . (11.5%) (6.0%) (2.3%) (2.8%)
t includes 201 subjects with HIV test interval >36 months
o Table 5.4 Age at seroconversion of all subjects reported to the Register! by exposure category and sex
Variable Age (years) Median 25th, 75th Mean s.d Range percentiles
Exposure category Sex between men 28 25 34 30.1 7.9 16- 70 Sex between men & women 29 24 35 30.5 8.9 17- 57 Injecting Drug Use 23 20 28 24.8 5.6 15-44 Other/undetermined 34 27 37 32.9 8.5 15-59 Sex Men 28 24 34 30.0 8.0 15-70 Women 25 22 31 27.1 7.7 15- 59
Overall 28 24 34 29.7 8.0 15- 70
t includes 201 subjects with HIV test interval >36 months
o Figure 5.3 Age distribution of seroconverters by estimated year of seroconversion
300
□ <20 yrs
250 □ 20-24 yrs □ 25-29 yrs □ 30-34 yrs 200 0 35-39 yrs □ 40-44 yrs □ 45-49 yrs 150 □ 50 VTS+
100
50 g
82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97
o all statistical analysis. Of the 1821 seroconverters with a test interval of 36 months or less, 1007 (55.3%) had seroconverted within one calendar year of their last negative antibody test (Table 5.5).
Seventy-six persons had a test interval of less than one calendar month. The proportion of these persons who have been diagnosed with AIDS and who have died is higher than in persons with wider test intervals. This may suggest that persons with a test interval of less than one calendar month are a peculiar group of individuals. The narrowness of this interval suggests that they may have been tested during the primary HIV infection stage. For 48 persons with less than one month test interval, information on seroconversion illness was recorded, and of these 34 (70.8%) reported an illness. This is highly significant in comparison with persons with a seroconversion interval of one to 36 months where this proportion was 28.2% (281 of 997,/) < 0.01).
5.7 AIDS events
AIDS was diagnosed in 494 seroconverters between 1986 and 1997 (Table 5.6). Twenty- two cases (4.5%) were diagnosed within one year of seroconversion. Of the 494 AIDS cases, information on the AIDS-defming condition was not reported for 23 persons. For the 471 persons with AIDS diseases reported, 22 had been diagnosed with 2 contemporaneously diagnosed conditions, and 3 with 3 contemporaneously diagnosed conditions. For 446 persons only one AIDS disease was present at the initial AIDS diagnosis.
For the 446 persons presenting with only one diagnosing disease, 230 had been definitively diagnosed and 79 were presumptively diagnosed (27 with PCP, 10 with KS, 26 with oesophageal candidiasis, and 16 with other conditions). For 137 persons the method of diagnosis was not reported. For the 25 persons presenting with more than one disease, 13 had at least one disease definitively diagnosed, and for 6 persons all diseases were presumptively diagnosed. For 6 other persons presenting with more than one disease, the method of diagnosis was not reported for any of them.
The most commonly presenting AIDS condition was PCP representing 30% (n=150) of initial AIDS diagnoses, followed by oesophageal candidiasis (n= 88) representing
108 Table 5.5 Interval between last negative and first positive HIV antibody tests and proportions with AIDS and who have died
Interval (months) AIDS (%) died (%) Totalf (100%)
< 1 22 (28.9) 16(21.1) 76
1-12 221 (23.7) 149 (16.0) 931 13- 24 135 (26.7) 86(17.0) 506 25- 36 75 (24.4) 47 (15.3) 308
37- 48 20 (22.5) 8 (9.0) 89 49 + 21 (18.8) 8(7.1) 112
Total 494 (24.4) 314(15.5) 2022 t includes 201 subjects with HIV test interval >36 months
109 Table 5.6 AIDS cases: Year of diagnosis of AIDS
Estimated year of Year of AIDS diagnosis Total seroconversion 86 87 88 89 90 91 92 93 94 95 96 97 1982 0 0 2 0 1 1 0 1 0 0 1 0 6 1983 0 0 0 5 1 1 5 3 4 1 1 0 21 1984 0 1 4 1 4 2 2 7 3 4 0 0 28 1985 0 3 0 5 3 5 1 4 . 1 0 4 0 27* 1986 2 1 2 1 9 4 5 8 4 1 1 0 38
1987 - 0 3 2 3 2 4 7 10 7 1 0 39
1988 - - 1 2 4 4 9 5 17 3 5 1 51
1989 --- 0 4 6 7 4 21 8 6 0 56
1990 - - -- 1 2 6 17 14 15 3 1 59
1991 -- - -- 4 10 12 11 19 7 0 63
1992 ------1 9 14 18 9 0 51
1993 ------6 7 12 9 0 34
1994 ------1 8 2 0 11
1995 ------5 3 1 9
1996 ------1 0 1 Total 2 5 12 16 30 31 50 83 107 101 53 3 494
* includes 1 AIDS date not known 17.6% of AIDS cases (Table 5.7). Two conditions introduced in the 1993 AIDS case definition (pulmonary tuberculosis and recurrent pneumonia) accounted for 5.4% (n= 27) of total AIDS diagnoses. The most striking difference in the distribution of diseases among the exposure categories is the diagnosis of KS (Table 5.8). Of 69 persons who presented with this condition, 64 were homosexual and bisexual men, representing 17.2% of AIDS diseases diagnosed in that exposure category. Three persons infected heterosexually were also reported with KS, 2 males aged 27 and 36 years (one of whom was a black African), and one 28 year old black African female. Both persons diagnosed with KS with undetermined exposure category were white males, aged 31 and 37 years. Further, there appeared to be a marginally significant excess of pulmonary TB, MAI and recurrent pneumonia diagnoses among persons infected heterosexually.
In the analysis of HIV seroconversion to AIDS 1528 persons were not diagnosed with AIDS, of whom 55 had died without AIDS (Table 5.9). Of the 1473 AIDS-free individuals who were not known to have died, 1197 (81.3%) had been clinically assessed in 1995 or later.
5.8 Death
By 30 September 1997, 314 seroconverters were known to have died between 1986 and 1997 (Table 5.10). Of these, 55 (17.5%) had not been diagnosed with AIDS. Four persons (1.3%) died within one year of seroconverting. The proportion of persons dying without an AIDS diagnosis was significantly higher among injecting drug users than among other exposure categories with 40.5% (32 of 79) IDUs dying without AIDS compared to 10.4% (20 of 193) of homosexual men, 8.3% (2 of 24) of those heterosexually infected, and 5.6% (1 of 18) of those infected by other or unknown means.
The cause(s) of death for persons who died without an AIDS diagnosis were given as follows: Overdose/suicide (13), chronic liver disease/hepatitis B (4), accidental (3), bronchopneumonia (3), drug dependency (2), meningitis (1), peritonitis (1), respiratory disease (1), abdominal lymphoma (1), and renal failure (1). The cause(s) of death was not reported for 25 individuals who died without AIDS.
Ill Table 5.7 AIDS cases: AIDS defining diseases and proportion of total diagnoses
Disease(s) Number Number diagnosed incl. diagnosed persons with > 1 disease* with one (% of total initial disease diagnoses)
Pneumocystis carinii pneumonia (PCP) 144 150 (30.0) Oesophageal candidiasis 76 88 (17.6) Kaposi's Sarcoma (KS) 67 69(13.8) HIV Encephalopathy 24 29 (5.8) HIV Wasting syndrome 18 23 (4.6) Herpes simplex infection 17 18(3.6) Cryptosporidiosis 15 16(3.2) Atypical mycobacterial infection (MAI), 13 14 (2.8) disseminated Pulmonary tuberculosis (PTB) 13 14(2.8) Pneumonia, recurrent within 12- mth period 12 13 (2.6) Extrapulmonary tuberculosis (EPTB) 8 10(2.0) Candidiasis, other invasive 6 10(2.0) Non-Hodgkin’s Lymphoma 8 9(1.8) CMV Retinitis 6 9(1.8) CMV disease, other 3 6(1.2) Progressive multi-focal leukoencephalopathy 5 5(1.0) Primary CNS lymphoma 3 5(1.0) Cerebral toxoplasmosis 2 4 (0.8) Extrapulmonary cryptocoeeal infection 2 3 (0.6) Atypical mycobacterial infection (not MAI) 2 2 (0.4) Isosporiasis 1 1 (0.2) Salmonella septieaemia/baeteraemia, recurrent 1 1 (0.2) Total 446 499 (100) Disease(s) not reported 23
* 25 persons were reported to have more than one AIDS-defining disease present at the initial AIDS diagnosis
112 Table 5.8 AIDS cases: exposure category distribution of all AIDS-defining conditions
Disease sex between men sex between men IDU other/ & women undetermined Pneumocystis carinii pneumonia 103 (27.7) 14 (30.4) 24 (38.7) 9 (47.4) Oesophageal candidiasis 69(18.5) 4 (8.7) 14 (22.6) 1 (5.3) Kaposi's Sarcoma 64(17.2) 3 (6.5) 0 2(10.5) HIV Encephalopathy 19(5.1) 4(8.7) 6(9.7) 0 HIV wasting syndrome 16(4.3) 4(8.7) 3 (4.8) 0 Herpes simplex infection 13 (3.5) 2(4.3) 2(3.2) 1 (5.3) Cryptosporidiosis 16(4.3) 0 0 0 Pulmonary tuberculosis (PTB) 6(1.6) 4(8.7) 1 (1.6) 3 (15.8) Atypical mycobacterial infection (MAI), disseminated 8 (2.2) 4(8.7) 1 (1.6) 1 (5.3) Pneumonia, recurrent within 12-mth period 8 (2.2) 3(6.5) 1 (1.6) 1 (5.3) Candidiasis, other invasive 9(2.4) 1 (2.2) 0 0 Extrapulmonary tuberculosis (EPTB) 9(2.4) 1 (2.2) 0 0 Non-Hodgkin’s Lymphoma 8 (2.2) 0 0 1 (5.3) CMV Retinitis 7(1.9) 1 (2.2) 1(1.6) 0 CMV disease, other 6(1.6) 0 0 0 Primary CNS lymphoma 4(1.1) 0 1 (1.6) 0 Progressive multi-focal leukoencephalopathy 3 (0.8) 0 • 2(3.2) 0 Cerebral toxoplasmosis 2 (0.5) 0 2(3.2) 0 Extrapulmonary cryptococcal infection 1 (0.3) 0 2 (3.2) 0 Atypical mycobacterial infection (not MAI) 0 0 2(3.2) 0 Isosporiasis 1 (0.3) 0 0 0 Salmonella septieaemia/baeteraemia, recurrent 0 1 (2.2) 0 0 Total (100%) 372 46 62 19
U) Table 5.9 Non-pregressors: year of last clinic visit for persons not diagnosed with AIDS
Estimated year of Year last assessed Total seroconversion 85 86 87 88 89 90 91 92 93 94 95 96 97 1982 0 0 0 0 0 0 0 0 0 0 0 1 0 1 1983 0 1 1 1 1 1 0 1 0 1 2 7 0 16 1984 0 1 0 0 0 0 1 0 1 1 1 15 0 20 1985 1 0 0 0 0 0 0 0 2 2 2 6 4 17
1986 - 3 2 0 2 0 0 1 2 0 3 7 10 30
1987 - - 2 0 1 2 1 1 4 0 5 14 7 37
1988 - -- 1 4 3 3 3 3 5 10 14 17 63
1989 -- -- 1 7 7 4 6 7 31 23 26 112
1990 ---- - 2 13 13 8 17 38 34 23 148
1991 ------4 17 11 15 54 45 20 166
1992 ------8 18 25 62 51 49 213
1993 ------7 22 79 77 56 241
1994 ------5 41 63 87 196
1995 ------19 66 58 143
1996 ------14 47 61
1997 ------9 9 Total alive 1 5 5 2 9 15 29 48 62 100 347 437 413 1473 (no. died without (3) (1) (1) (7) (2) (7) (7) (8) (11) (8) (55) AIDS) Table 5.10 Deaths: Estimated year of seroconversion by year of death
Estimated year of Year of death Total seroconversion 86 87 88 89 90 91 92 93 94 95 96 97 1982 0 0 0 0 1 2 1 0 1 0 0 0 5 1983 0 1 0 1 1 4 4 3 5 5 3 0 27 1984 0 1 1 2 8 2 1 3 7 7 0 0 32 1985 0 2 0 3 0 4 3 5 2 0 0 0 19 1986 1 0 2 1 4 5 5 5 10 3 1 0 37
1987 - 0 0 0 3 1 2 5 6 5 5 1 28
1988 - - 0 2 1 3 3 7 11 5 5 0 37
1989 - -- 0 1 1 3 6 6 11 4 0 32
1990 - -- - 1 1 3 4 11 7 9 0 36
1991 -- -- - 0 2 4 6 9 6 0 27
1992 ------0 0 3 7 10 0 20
1993 ------1 3 2 3 0 9
1994 ------0 1 2 0 3
1995 ------1 1 0 2 Total 1 4 3 9 20 23 27 43 71 63 49 1 314 In total, 1708 seroconverters were not known to have died by 30 September 1997 (Table 5.11). Of these 1475 (86.4%) were known to be alive in 1995 or later.
5.9 Biological specimens stored
A total of 377 10 ml specimens of clotted blood have been collected for storage of serum from 299 seroconverters. Further 30 ml samples of bloods (20 ml EDTA, and 10 ml clotted) were also collected from 60 individuals within 18 months of seroconverting. Plasma and cells have been separated from these specimens and subsequently stored within the MRC’s Molecular Epidemiology Repositories in Cambridge and Edinburgh.
116 Table 5.11 Persons remaining alive in HIV survival estimates: year last known to be alive based on information from clinic
Estimated year of Year last known to be alive Total seroconversion 85 86 87 88 89 90 91 92 93 94 95 96 97 1982 0 0 0 0 0 0 0 0 0 0 0 2 0 2 1983 0 1 1 0 0 0 0 1 0 1 4 10 0 18 1984 0 1 0 1 0 0 0 0 1 1 1 20 0 25 1985 1 0 0 0 0 0 0 1 2 3 2 12 6 27
1986 - 2 3 0 0 0 0 2 2 0 4 9 12 34
1987 - - 2 0 0 2 1 0 4 2 11 20 8 50
1988 - -- 1 3 0 2 3 3 4 18 19 27 80
1989 - - - - 1 5 6 2 6 10 38 39 33 140
1990 - - --- 2 10 7 8 16 56 46 34 179
1991 ------4 10 10 15 74 61 33 207
1992 ------7 14 18 83 67 62 251
1993 ------6 21 84 91 66 268 1994 4 42 68 91 205 1995 18 71 62 151
1996 - 14 48 62 1997 9 9 Total 1 4 6 2 4 9 23 33 56 95 435 549 491 1708
o CHAPTER VI
STATISTICAL METHODS AND AN EVALUATION OF BIAS
This chapter details the methods used in the statistical analysis and explores the effect of a number of potential sources of bias previously highlighted in Chapter IV.
Progression estimates from HIV seroconversion to two end-points are estimated in this thesis: • AIDS • Death
For ease of terminology, the term ‘survival’ is used in this section to refer to a person remaining event-free, i.e. remaining AIDS-free when deriving AIDS progression estimates, and remaining alive when deriving survival estimates. An event is an uncensored observation and is termed a ‘failure’.
In estimating times to AIDS, AIDS was defined according to the case definition operable at the time of the diagnosis (see Chapter I). For example, a person diagnosed in 1990 with pulmonary TB (a disease added to the case definition in 1993) was not considered to have AIDS, unless another disease included in the 1987 definition was diagnosed at the same time. Persons who died without being diagnosed with AIDS were censored at the time of death.
As the data analysed in this thesis contain late entry and are censored, the statistical methods used were chosen as they allow for this. The choice of right censoring for each end point is detailed in section 6.7 with progression to AIDS being censored in June 1995, and progression to death censored in December 1996.
6.1 Kaplan-Meier
The survival fimction represents the probability that the event has not yet occurred. The Kaplan-Meier (Product-limit) method was used to estimate survival by dividing time since
118 seroconversion into discreet time intervals such that each interval ends at the time of an observation, censored or uncensored The probability of survival was then calculated at the end of each interval, and the product-limit survival function is represented as a step function which changes at each point associated with an event (an uncensored observation). The resulting curves were then used to visually display the survival function, and conventionally start at a probability of one at time zero (seroconversion). The curves were then used to examine whether the probability of failure is changing with follow up time, and for calculating survival probabilities for different periods of follow up.
6.2 Log-rank test
The log-rank test was used to test the effects of each covariate on progression rates for statistical significance. This was done by creating a sequence of (k x 2) contingency tables (covariate of interest X event observed/ event not observed at that time) one at each observed event time. A statistic based on the observed and expected values for these contingency tables is then calculated. The log-rank test was used initially to test the univariate effect of each covariate in turn on progression estimates.
6.3 Cox proportional hazard models
Cox proportional hazards models were then used to assess the independent effects of all covariates on progression times This was done by initially fitting all covariates of interest (the main effects), excluding one covariate, then comparing both models to examine whether the exclusion of this covariate resulted in a substantially worse fit (as given by the difference in -2 log likelihood distributed on Chi-square with degrees of freedom (df) equal to the difference in df between the two models).
This method also provides estimates of the relative hazard of progressing to each end point for each covariate of interest. Hazard rates for each category of covariates examined were then obtained (relative hazard= exp(P), where p= regression coefficient of covariate). The hazard function at time T is the probability that a subject will fail at time T, given that the subject has remained event-free to time T. The cumulative hazard function is the integral over time of the hazard function and is estimated as the negative logarithm of the survival function.
119 95% confidence intervals for the hazard ratio are given by:
exp (P± 1.96 X standard error(p)}
As this applies only if the relative hazard is constant, tests for non-proportionality were performed by first including an interaction term between time and the covariate of interest. If the inclusion of this term resulted in a significant improvement in fit, this was taken as evidence of non-proportionality.
6.4 Poisson regression
In order to examine the shape of the hazard at each year since seroconversion, and to assess whether the shape of the hazard has changed over calendar time, rates were calculated for each year since seroconversion as the number of events divided by the number of person years at risk during time t. A person’s time at risk is the time under follow up given that the person has not experienced the event by that time. These rates were then plotted.
Using a Poisson regression analysis relative rates were calculated together with 95% confidence intervals. The year of seroconversion was examined by including it in the regression model as a categorical variable (seroconverted 1983 to 1989, and seroconverted 1990 to 1995).
Kaplan-Meier, log rank, Cox models and Poisson analysis were carried out using the ‘st’ commands of the statistical package ST AT A version 5.0 This command declares data to be survival data with a variable containing individual time of entry into the risk set (otherwise it is assumed to be from seroconversion).
6.5 Documentation of seroconversion
For 1281 seroconverters, 63.4% of total reported, documentation of the negative test date was possible of whom 1202 persons fulfilled the Register’s criteria of a 3-year maximum HIV test interval.
120 The relatively high proportion of documented seroconverters among the injecting drug users (87.0%) is due to the fact that most IDUs reported to the Register were from one well-defined cohort in Edinburgh (127 of 167 documented IDUs with a 3-year maximum HIV test interval) (Table 6.1). The Edinburgh Hospital Clinic cohort is described in Chapter II.
Seroconverters for whom the documentation of the antibody negative test was not verified were found to progress to AIDS significantly more rapidly than documented seroconverters (RR= 1.8, 95% CI= 1.3- 2.3). Figure 6.1 shows Kaplan-Meier plots for documented and undocumented seroconverters. Progression estimates are in the order of 24 months difference between the two groups suggesting that, on average, undocumented seroconverters may have tested antibody negative some 24 months earlier than reported.
Due to the size of this discrepancy, all further analyses were restricted to the 1202 persons with a documented negative HIV test with a maximum HIV test interval of 36 months.
6.6 Late entry
Due to the largely retrospective method by which seroconverters have been identified for inclusion in the Register, a number of subjects were treated as late entry subjects in the progression analysis as they did not come under unbiased follow-up until some time after seroconversion, (see Chapter IV- section 4.5: Left truncation of survival times from seroconversion).
Of 1202 seroconverters reported to the Register with a maximum test interval of 36 months, 285 did not contribute any information to progression estimates from HIV seroconversion to AIDS as they seroconverted after 30 June 1995, or entered the risk set on or after either: the date they were last clinically assessed (if censored), or the date of AIDS diagnosis (for those with AIDS). The remaining 917 seroconverters contributed information to estimates of the AIDS incubation period, 543 of whom contribute from seroconversion and 374 were treated as late entry subjects. These 374 late entry subjects contributed information to progression estimates from anything between 14 days following seroconversion to 11.1 years following seroconversion.
121 Table 6.1
Documentation of seroconversion by exposure category for seroconverters with a 3-year maximum HIV test interval and proportion of total reported
Exposure category Documented seroconverters (% of total reported for this exposure group)
Sex between men 906 (58.0%) Sex between men and women 114(52.8%) Injecting drug use 167 (87.0%) Other/undetermined 15 (29.4%)
Total 1202 (59.4%)
Figure 6.1
Kaplan-Meier plots of time to AIDS from HIV seroconversion for documented and undocumented seroconverters
1.00
0.75 docum ented seroconverters
0.50
undocumented seroconverters
0.25
0.00
4 5 6 7 8 9 10 11 12 13 14
Time from seroconversion (yrs)
122 For the analysis of estimates from seroconversion to death, 53 individuals did not contribute any information to progression estimates as they seroconverted after 31 December 1996, or entered the risk set on or after either: the date they were last known to be alive (if censored), or the date of death. The remaining 1149 seroconverters contributed information to progression estimates, 589 of whom contributed from seroconversion and 560 were treated as late entry subjects.
A summary of characteristics of the 1149 seroconverters in the survival analysis and 917 in the AIDS estimates are given in Tables 6.2 and 6.3.
In order to examine the effect of late entry on AIDS progression estimates, Kaplan- Meier estimates were made allowing for late entry, and repeated ignoring late entry and the resulting plots compared.
In this exercise, late entry was found to have little effect on progression estimates (Figure 6.2). Ignoring late entry tended to produce slightly slower progression estimates. As expected, ignoring late entry increased the number of persons in the risk set particularly more so in the first 3 years following seroconversion which reduced the width of the confidence intervals around the point estimates (Table 6.4).
The increase in progression rates after allowing for late entry was expected. However, the size of this reduction was surprising and it may be that the choice of date of entry into the risk set for seroconverters from each centre was too conservative (i.e. allowing seroconverters to enter the risk set later than would he plausible).
All further analysis, however, allowed for late entry.
6.7 Right censoring strategies
A number of right censoring strategies were used in estimating progression rates from HIV seroconversion to AIDS:
• strategy 1- subjects were censored on the date they were last known to he AIDS-free, and persons who died without AIDS were censored on the date of death.
123 Table 6.2
Characteristics of 917 documented seroconverters with 3 year maximum test interval and contributing information to the risk set of AIDS analysis
Number of Proportion reports of cases (n=917) (%) Sex Male 805 87.8 Female 112 12.2 Exposure category Sex between men 671 73.2 Sex between men & women 80 8.7 Injecting drug use 156 17.0 Other/undetermined 10 1.1 HIV test interval within 1 calendar month 54 5.9 1-12 months 478 52.1 13-24 months 248 27.0 25-36 months 137 14.9
How ascertained Clinic 595 64.9 Laboratory and/or clinic 322 35.1 Seroconversion Yes 141 15.4 illness reported No 283 30.9
Not known 493 53.8 Age at Median 27 years (range= 16-70) seroconversion
124 Table 6.3 Characteristics of 1149 documented seroconverters with 3 year maximum test interval contributing information to the risk set of HIV survival analysis
Number of Proportion reports of cases (n=1149) (%) Sex Male 1013 8&2 Female 136 11.8 Exposure category Sex between men 860 74.8 Sex between men & women 109 9.5 Injecting drug use 166 14.4 Other/undetermined 14 1.2 HIV test interval within 1 calendar month 68 5.9 1-12 months 613 53.4 13-24 months 298 25.9 25-36 months 170 14.8 How ascertained Clinic 734 6 3 9 Laboratory and/or clinic 415 36.1 Seroconversion Yes 204 17.8 illness reported No 405 35.2 Not known 540 47.0 Age at Median 28 years (range= 16- 70) seroconversion
125 Figure 6.2
Kaplan-Meier plots of time to AIDS from HIV seroconversion allowing for late entry and ignoring it
1.00
0.75 ignoring late entry
£ a. 0.50 I allowing for Q l a t e e n t r y
0.25
0.00
1 2 3 4 5 6 7 8 9 10 11 12
Time from seroconversion (j rs)
Table 6.4 Examination of the effect of ignoring late entry on AIDS progression estimates
Time from allowing for late entry ignoring late entry seroconversion( proportion s.d. number in proportion s.d. number in yrs) AIDS-free risk set AIDS-free risk set 1 0982 0.0057 537 0986 0.0036 989 2 0.954 0.0089 509 0.957 0.0065 840 3 0937 0.0105 443 0.936 0.0081 684 4 0862 0.0158 359 0.875 0.0121 522 5 0.804 0.0188 287 0.818 0.0150 387 6 0.729 0.0224 214 0.746 0.0186 278 7 0.635 0.0264 152 0.652 0.0229 189 8 0.580 0.0284 116 0.588 0.0254 137 9 0.510 0.0314 81 0.521 0.0285 93 10 0.453 0 0339 56 0.466 0.0315 60 11 0362 0.0384 31 0.379 0.0368 33 12 0.350 0.0391 5 0.367 0.0376 5
126 • strategy 2 (preferred option)- subjects were uniformly censored on 30 June 1995, with those without AIDS being regarded as AIDS-free on that date. Subjects with AIDS diagnosed after that date were regarded as AIDS-free, and persons who died after that date were regarded as alive. Subjects who died without AIDS were censored on the date of death. Subjects who seroconverted after that date were excluded from the analysis.
• strategy 3- subjects were uniformly censored on 31 December 1996, with those without AIDS being regarded as AIDS-free on that date. Subjects with AIDS diagnosed after that date were regarded as AIDS-free, and persons who died after that date were regarded as alive. Subjects who died without AIDS were censored on the date of death. Subjects who seroconverted after that date were excluded from the analysis.
• strategy 4- subjects were uniformly censored on 30 April 1997, with those without AIDS being regarded as AIDS-free on that date. Subjects with AIDS diagnosed after that date were regarded as AIDS-free, and persons who died after that date were regarded as alive. Subjects who died without AIDS were censored on the date of death. Subjects who seroconverted after that date were excluded from the analysis.
• strategy 5- subjects were uniformly censored on 30 September 1994, with those without AIDS being regarded as AIDS-free on that date. Subjects with AIDS diagnosed after that date were regarded as AIDS-free, and persons who died after that date were regarded as alive. Subjects who died without AIDS were censored on the date of death. Subjects who seroconverted after that date were excluded from the analysis. For survival estimates, three alternate right censoring strategies were explored:
• strategy 1- subjects who have not died were censored on the date they were last known to be alive.
• strategy 2 (preferred option)- subjects last seen before 1 January 1993 were censored on the date they were last seen. Subjects from clinics in England and Wales last seen between 1 January 1993 and 31 December 1996 (death information included in the ONS file) were uniformly censored on 31 December 1996 with persons not known to have died being regarded as alive on 31 December 1996. All subjects from clinics in
127 Scotland not reported to have died were censored on the date they were last seen. Subjects who died after 31 December 1996 were regarded as alive. Subjects who seroconverted after 31 December 1996 were excluded from the analysis.
• strategy 3- subjects were uniformly censored on 30 September 1994 with persons not known to have died being regarded as alive on that date. Subjects who died after that date were regarded as alive. Subjects who seroconverted after that date were excluded from the analysis.
As illustrated by the Kaplan-Meier curves for progression to AIDS in Figure 6.3, censoring according to strategy 1 resulted in a short time to AIDS from seroconversion with an estimated median time to AIDS of 8.14 years (95% CI= 7.28- 8.83 years). Each of strategies 2, 3 and 4 produced increasingly longer times to AIDS with median times from seroconversion to AIDS of 9.26, 9.72 and 10.78 years respectively as increasingly more optimistic assumptions were being made relating to subjects remaining AIDS-free. Under strategy 5, with subjects censored on 30 September 1994 (a valid assumption as few subjects were truly not seen in the clinic prior to that date), a median estimate of time to AIDS of 9.26 years (95% CI= 8.54- 10.58 years) was derived, which is close to estimates derived under strategy 2.
Having established the AIDS status of seroconverters through other registries these events will be over-represented unless persons not found on those registries are treated likewise, i.e. regarded as being event-free at that time.
For survival estimates, strategy 1 produced the shortest time to death with a median survival time of 10.09 years (95% CI= 9.24- 10.95 years) (Figure 6.4). Under strategy 2, a median time of 10.79 years (95% CI= 9.81- 11.36) was obtained, and 10.58 years (95% CI= 9.30- 11.74) under strategy 3 (censored on 30 September 1994). This may suggest that strategy 2 may be slightly over-optimistic in assuming persons were still alive who may, in fact, have died.
In the following chapter, results from progression estimates to AIDS and death are presented. For AIDS estimates strategy 2 was used (censored on 30 June 1995) and for survival estimates, strategy 2 was used (censored on 31 December 1996).
128 Figure 6.3
Kaplan-Meier plots of the effect of alternate right censoring strategies on AIDS progression estimates
1.00
.50
strategy
strategy 2
strategy 3
strategy 4
strategy 5 .00
1 2 3 4 5 6 7 8 9 10 11 12 13 14
Time from seroconversion (>rs)
129 Figure 6.4
Kaplan-Meier plots of the effect of alternate right censoring strategies on HIV survival estimates
1 .00
s t r a t e g y 3
s t r a t e g y 2 a a a« I ■3> 0.50 _ t s t r a t e g y 1 a
0.00 -
1 2 3 4 5 6 7 8 9 10 11 12 13 14
Time from seroconversion (> rs)
130 CHAPTER VII
PROGRESSION ESTIMATES FROM HIV SEROCONVERSION TO AIDS AND TO DEATH
Of 1202 documented seroconverters with a maximum HIV test interval of 3 years, 917 and 1149 seroconverters contributed information to progression estimates from HIV seroconversion to AIDS and to death (from any cause) respectively (Section 6.6). A flow chart and table summarising the numbers of subjects reported to the Register and the numbers included in the final analyses are given in Figure 7.1 and Table 7.1.
Two annual follow ups of seroconverters reported to the Register were undertaken in February 1996 and March 1997, providing the most recent information on the clinical state and vital status of seroconverters. Identifying information of persons not reported by the clinical centre to have AIDS or to have died were subsequently matched against the national HIV and AIDS surveillance databases at CDSC and SCIEH in June 1997 (Section 3.6.1.3). Allowing a 2-year reporting delay persons not found on the AIDS registers at CDSC and SCIEH were assumed to be AIDS-free in June 1995.
Persons who died without an AIDS diagnosis were censored on the date of death, otherwise all those AIDS-free were censored in June 1995.
Further, a cross check was made using the soundex code, date of birth and sex of seroconverters with the ONS data on persons who have died in England and Wales between 1 January 1993 and 31 December 1996 (see section 3.6.1.4). Persons not known to have died in Scotland and Northern Ireland were censored at the date they were last seen, as were all persons last seen before 1 January 1993. Otherwise, persons not reported to have died were assumed to be alive in December 1996.
7.1 Examining the effects of covariates on AIDS progression estimates
By June 1995, 181 persons (19.7%) had been diagnosed with AIDS. The overall Kaplan- Meier cumulative progression estimates to AIDS from seroconversion are given in Table
131 Figure 7.1 A flow chart of number of subjects reported to the Register and numbers included in final analyses
2022 HIV test interval persons > 36 months reported
HIV test interval < 36 months 201
undocumented 1821 negative test
documented negative test
619 1202
contributing information, and therefore included in analyses
AIDS Death analysis analysis
from a time after from a time after seroconversion 917 1149 seroconversion
from seroconversion from seroconversion
374 543 589 560
132 Table 7.1 A descriptive comparison between; ; N- the total numbers of subjects reported to the Register (n= 2022) and the numbers included in the AIDS (n= 917) (A) and death (n= 1149) (D) progression analyses
NAD (n=2022) (n=917) (n=1149)
Sex Male 1816 805 1013 (90%) (88%) (88%) Female 206 112 136 Exposure Sex between men 1563 671 860 category (77%) (73%) (75%)
Sex between men & women 216(11%) 80 (9%) 109 (10%) Injecting drug use 192(10%) 156(17%) 166 (14%) Other/undetermined 51 (3%) 10(1%) 14(1%) HIV test within 1 calendar month 76 (4%) 54(6%) 68 (6%) interval 1-12 months 931 (46%) 478 (52%) 613 (53%) 13-24 months 506 (25%) 248 (27%) 298 (26%) 25-36 months 308 (15%) 137 (15%) 170 (15%) >36 months 201 (10%) -- How Clinic 1511 595 734 ascertained (75%) (65%) (64%) Laboratory and/or clinic 511 322 415
Seroconversion Yes 325 (16%) 141 (15%) 204 (18%) illness reported No 756 (37%) 283 (31%) 405 (35%) Not known 941 (47%) 493 (54%) 540 (47%) Age at seroconversion 28 27 28 Median (range) years (15- 70) (16- 70) (16- 70)
133 12 giving a median time to AIDS of 9.26 years (95% confidence intervals^ 8.46- 10.39 years) for a population with a similar age profile (Figure 7.2).
Table 7.3 summarises the effect of covariates examined using Cox analyses on AIDS progression estimates. In univariate analysis, the following covariates were found to be significantly associated with progression from HIV seroconversion to AIDS: • Exposure category. Persons exposed through sex between men and women (RR= 0.51, 95% confidence intervals= 0.26- 0.97) as well as injecting drug users (RR= 0.53, 95% confidence intervals= 0.36- 0.79) were found to progress more slowly than persons exposed through sex between men. • Sex. Women were found to progress more slowly than men (RR= 0.46, 95% confidence intervals^ 0.28- 0.74). • Age. Younger individuals were found to progress more slowly than older subjects with an estimated relative risk of 1.05 (95% confidence intervals=1.03- 1.06) for each one year increase of age at the time of HIV seroconversion. • HIV test interval. Persons with an interval between one and 36 months were found to progress more slowly than those diagnosed positive within one calendar month of the negative test. The latter will be referred to as ‘acute’ infections. • Year of seroconversion. A slight increase in the risk of developing AIDS by year of seroconversion was observed. On examining this further by fitting year of seroconversion as a categorical variable no evidence of a trend was observed.
In order to examine these factors further, Cox models were fitted for each covariate of interest whilst adjusting for the effect of: age, sex, exposure category and HIV test interval.
7.1.1 Exposure category
No evidence of a significant difference between exposure categories was observed. In univariate analysis, compared to men who have sex with men, male IDUs appeared to progress to AIDS more slowly with a relative risk of 0.61 (95% CI= 0.39- 0.95). After adjusting for age, this increased and became non-significant at the 5% level to 0.70 (95% CI= 0.45- 1.09). Compared to men who have sex with men, men exposed heterosexually had an increased non-significant risk of being diagnosed with AIDS with
134 Table 7.2 Cumulative proportion of seroconverters developing AIDS
Time from Cumulative proportion with number in number seroconversion AIDS (95% CI) risk set with AIDS (yrs) 1 1.8 (1.0- 3.4) 537 10
2 4.6 (3.1- 6.7) 509 15
3 6.3 (4.6- 8.7) 443 9
4 13.8 (11.0-17.2) 359 33
5 19.6 (16.2- 23.6) 287 23
6 27.1 (22.9- 31.8) 214 24
7 36.5(31.6-41.9) 152 25
8 42.0 (36.6- 47.8) 116 12
9 49.0 (43.0- 55.3) 81 12
10 54.8 (48.3- 61.5) 56 8
11 63.8 (56.3- 71.3) 31 9
12 65.0 (57.4- 72.6) 5 1
135 Figure 7.2
Cumulative progression rates from HIV seroconversion to an AIDS diagnosis
for seroconverters reported to the Register and censored in June 1995
with persons lost to follow up assumed to be AIDS-free on that date
1 . 00
75 o 15
0.50
0.25
0 . 00
4 5 6 7 8 10 11
Time from seroconversion (yrs)
136 Table 7.3 Factors associated with progression from HIV seroconversion to an AIDS diagnosis
Variable Relative Risk (95% Cl)
Univariate Multivariate*
Exposure Sex between men 1.00 1.00 category Sex between men & women 0.51 (0.26, 0.97) 0.78 (0.37, 1.64) Injecting drug users 0.53 (0.36, 0.79) 0.73 (0.48, 1.11) Others 0.96 (0.30, 3.02) p=0.004 1.02 (0.32,3.25) p=0.50
Sex Men 1.00 1.00 Women 0.46 (0.28, 0.74) p=0.0004 0.62 (0.35, 1.09) p=0.09
Age For each year 1.05 (1.03, 1.06) p=0.0001 1.04(1.02, 1.05) p=0.0002
How identified Clinic 1.00 1.00 Laboratory 1.20 (0.85, 1.48) p=0.31 1.08 (0.76, 1.55) p=0.66
HIV test interval same calendar mth 1.00 1.00 1-36 mths 0.50 (0.29,0.85) p=0.02 0.63 (0.36, 1.08) p=0.11
Calendar year of For each 1.07(1.00,1.15) p=0.06 1.02 (0.94, 1.11) p=0.57 seroconversion subsequent year
* Adjusted for exposure category, sex, age at seroconversion, and HIV test interval w relative risk of 1.46 (95% CI= 0.64- 3.31). After adjusting for age this was reduced to a relative risk of 1.27 (95% CI= 0.55- 2.90). The inclusion of acute seroconversion as a covariate had little influence on these estimates.
7.1.2 Sex
Women appeared to have a reduced risk of an AIDS diagnosis of 0.46 (95% CI= 0.28- 0.74) compared to men. Adjustment for age caused a slight modification, with women still progressing significantly more slowly than men (RR= 0.53, 95% CI= 0.32- 0.86), and similarly when only the effect of exposure category was adjusted for (RR= 0.55, 95% CI= 0.31- 0.99). The effect between men and women was not found to be significant, however, when the effects of both age and exposure category were adjusted for (RR= 0.62, 95% CI= 0.35- 1.09).
7.1.3 Age
At the 5% level, age was found to be the only factor independently significantly associated with progression to AIDS. Models using quadratic and cubic terms for age did not improve the fit. After adjusting for exposure category, sex and HIV test interval, a relative risk of 1.036 per one year increase in age was estimated (95% CI= 1.018- 1.055). This is equivalent to a 43% increase in risk for a 10 year increase in age (95% CI= 19-71%).
Five years following seroconversion, progression to AIDS was estimated to be 13.5% (95% CI= 8.9- 20.2), 19.5% (95% CI= 14.6- 25.8), 23.2% (95% C l- 15.0- 34.8), and 47.0% (95% CI= 29.4- 68.5) respectively for age groups 16- 24, 25- 34, 35- 44, and 45 years or over (Figure 7.3). Median times (and 95% Cl) from seroconversion to AIDS for age groups 16-24, 25-34, and 35-44 years were estimated to be 10.78 (9.58- oo), 9.19 (7.41- 10.75) and 6.8 (6.27- 9.36) respectively. The median time for persons aged 45 years or over could not be estimated due to few persons remaining in the risk set beyond 5 years from seroconversion. The relative risk of AIDS was estimated to be 0.68 (95% Cl- 0.48- 0.96) for persons aged under 25 years compared with persons aged 25- 34 years. For persons aged 35- 44 years and those aged 45 years or over the relative risk
138 Figure 7.3
Kaplan-Meier plots for time to AIDS from HIV seroconversion by age group at HIV seroconversion
. 00
16- 24 yrs
.75 £
25- 34 y: 2 s. .50 c/jI 9 4 5 yrs or over <
35- 44 yrs .25 -
0.00
1 2 3 4 5 6 7 8 9 10 11
Time from seroconversion (yrs)
Table 7.4 Examination of the effect of age at seroconversion on progression estimates from HIV seroconversion to AIDS
Age group at RR* 95% seroconversion (yrs) confidence intervals
16- 24 068 0.48- 0.96
25- 34 1
35-44 1.43 0.95- 2.15
45 and over 3.09 1.74-5.50
Log likelihood = 24.59,p « 0.001 (not adjusted for other variables)
139 was higher (1.43, 95% CI= 0.95- 2.15, and 3.09, 95% CI= 1.74- 5.50) respectively Table 7.4). There was also a suggestion that the risk of AIDS associated with a one year increase in age is greater for persons aged 45 years or over than the risk associated with a one year increase in persons under the age of 45. I was unable to evaluate this statistically, however, as only 50 seroconverters (14 with AIDS) were aged 45 years or over.
Examining the age effect for men exposed through sex between men, a strong independently significant effect was observed with an estimated relative risk of 1.04 (95% CI= 1.02- 1.06) per one year increase in age (Figure 7.4).
For men exposed through sex between men the relative risk of AIDS for those aged 45 years or over at seroconversion was 2.66 (95% CI= 1.41- 5.00) compared to those aged 25- 34 years. The effect of age was broadly similar to that observed for the cohort as a whole (Table 7.5).
7.1.4 How identified
No evidence of a difference in risk between seroconverters identified through laboratory and clinic records was found on univariate and multivariate analysis. Persons identified through the laboratory having a relative risk of AIDS of 1.08 (95% CI= 0.76- 1.55) after adjusting for exposure category, sex, age, and HIV test interval (Table 7.3). In any case, it may not be feasible to disentangle the method of ascertainment of seroconverters as laboratory information on the presence of a previous negative antibody test may have come via the clinical centre when the request (for the positive test) was made. It may be impossible, therefore, to determine which is the primary source of information.
7.1.5 HIV test interval
Persons testing positive within the same calendar month as the negative test were found to have significantly faster progression to AIDS on univariate analysis and only marginally significant in multivariate analysis (p=0.11) (Table 7.3). This was explored further to examine whether this effect persisted throughout the time of follow up. The finding appeared to be almost entirely due to 3 cases (twohomosexual men aged 26 and
140 Figure 7.4
Kaplan-Meier plots for time to AIDS from HIV seroconversion
by age group at HIV seroconversion for men exposed through sex between men
1.00 -
1 6 - 24 y r s
Q
2 5 - 34 4 5 yrs or over
0.50
3 5 - y r s
0.25
0.00
1 3 4 5 6 7 8 9
Time from seroconversion (\rs)
Table 7.5 Examining the effect of age at seroconversion on AIDS progression for men exposed through sex between men
Age group (yrs) RR Cumulative progression to
(95% Cl)* AIDS 5 years following
seroconversion (95% Cl)
16- 24 0.88 (0.56- 1.40) 19.5 (12.3-30.0)
25- 34 1 21.7 (15.9-29.4)
35- 44 1.59 (0.99-2.55) 25.4 (16.0-38.8)
45 or over 2.66 (1.41-5.00) 45.5 (27.4-68.4)
* Log likelihood x^3 ^ 11 98, p= 0.0075 (not adjusted for other variables)
141 34 years, and one heterosexual man aged 48 years) who presented with AIDS within 3 months of HIV seroconversion in 1992 and 1993 and for whom seroconversion was estimated within one calendar month and in whom AIDS developed within 17, 25, 74 days from estimated seroconversion.
7.1.6 Calendar time
To examine the hazard at each year following seroconversion, event rates and 95% confidence intervals were calculated and plotted by time since seroconversion (Figure 7.5). The rate appears to be low in the first three years following seroconversion (equivalent to 18- 28 events per 1000 person-years at risk) after which it increases slightly and plateaus at around 8 years following seroconversion (91 events per 1000 person-years at risk). This study, however, is under-powered to properly assess these changes over time. Further, from these data, there was no evidence that the hazard has changed for persons who seroconverted before 1989 compared with persons who seroconverted in 1989 or later (x^i = 0.0149,/?= 0.90).
The calendar year of seroconversion was found to be marginally significantly associated with progression in univariate analysis (p= 0.06), with a slightly increased hazard in more recent time periods (i.e. time to AIDS from seroconversion shortening). This finding, however, appeared to be due to more IDUs (who are younger and with a lower hazard) seroconverting in the earlier time periods. After adjusting for exposure category, the relative hazard was estimated to be 1.02 (95% CI= 0.94- 1.11).
Examination of calendar year of seroconversion for men exposed through sex between men found no evidence of a change over time in this group of seroconverters.
7.2 Width of HIV test interval
Restriction of analysis to seroconverters with narrower HIV test intervals appeared to have little effect on AIDS progression estimates overall and within the first 3 years of seroconversion (Figure 7.6 & Table 7.6). There was, however, a slight tendency for higher estimates from persons with narrower test intervals. This may be due to the higher
142 Figure 7.5
Estimated AIDS rate and 95% confidence intervals
for each year since seroconversion
400
s %© ©k. CL § O 200 1 « 100 2 (/3 Q
I I 5 6 7 8 9 10 1 1 12 13
Time from seroconversion (yrs)
143 Figure 7.6
Kaplan-Meier plots of the effect of the HIV test interval on progression estimates from HIV seroconversion to AIDS
1.00
.75
.50 5 o <
.25 maximum interval -12 months
maximum interval- 24 months
maximum mterval -36 months
.00
4 5 6 7 10 11
Time from seroconversion (yrs)
Table 7.6 The effect of narrower HIV test intervals on AIDS progression estimates within the first 3 years of HIV seroconversion
Time from Cumulative progression estimates to AIDS (95% Cl) seroconversion Seroconverters with Seroconverters with Seroconverters with
(yrs) 3- yr maximum 2- yr maximum 1- yr maximum interval interval interval
1 1.8 (1.0-3.4) 1.9 (1.0-3.6) 2.3(12- 4.6) 2 4.6 (3.1-6.7) 4.6 (3.1-7.0) 5.2(3.2-82) 3 6.3 (4.6- 8.7) 6.5 (4.6- 9.2) 6.8 (4.6- 10.2)
144 proportion in these groups of persons with acute infection (testing positive in the same calendar month as the negative test), who were found to have higher progression rates. The inclusion criteria of persons with a maximum HIV test interval of 3 years does not appear to be unreasonable for deriving progression estimates.
7.3 Kaposi’s sarcoma
In order to examine the effect of a diagnosis of KS on the risk of developing AIDS, 19 homosexual men in the dataset diagnosed with KS were censored at the date of the KS diagnosis.
In univariate analysis of the effect of exposure category, male IDUs were found to have a non-significant reduced hazard of 0.75 (95% CI= 0.47- 1.17) compared to men who have sex with men. After adjusting for the effects of age, the relative hazard of developing non- KS AIDS for male IDUs was found to increase to 0.86 (95% CI= 0.54- 1.36).
7.4 Non-AIDS mortality
The effect of non- AIDS mortality, more common in IDUs and therefore leading to a greater proportion of cases censored in analysis to AIDS, was examined. Time was measured to the diagnosis of AIDS or death without AIDS.
After adjustment for age, there was no evidence of a risk difference among males, with male IDUs having a relative hazard of 0.99 (95% CI= 0.67- 1.45), and males exposed through sex between men and women having a relative hazard of 1.13 (95% CI= 0.50- 2.58) compared to men who have sex with men.
7.5 Examining the effects of covariates on HIV survival
By 31 December 1996, 203 seroconverters (17.7%) were known to have died. Kaplan- Meier estimates of cumulative progression to death from HIV seroconversion are given in Table 7.7 giving an estimated median time to death from any cause of 10.79 years (95% CI= 9.81-11.36 years) (Figure 7.7).
145 Figure 7.7
Cumulative progression rates from HIV seroconversion to death
for seroconverters reported to the Register and censored in December 1996
with persons lost to follow up assumed to be alive on that date
1 . 00
X! X« p c. .50 Î8
3 1/3
.25
. 00
1 2 3 4 5 6 7 8 9 10 11 12
Time from seroconversion (vrs)
146 Table 7.7 Cumulative proportion of seroconverters dying from any cause following HIV seroconversion
Time from Cumulative proportion number in number seroconversion dead (95% Cl) risk set dead
(yrs) 1 0.9 (0.4- 2.1) 655 6 2 2.6 (1.6- 4.1) 657 11 3 3.5 (2.3- 5.2) 630 6 4 7.0 (5.3- 9.2) 546 22
5 11.1 (8.8-13.9) 443 22 6 17.4 (14.4- 20.9) 358 29 7 24.7 (21.1- 28.9) 262 29 8 30.3 (26.1- 35.0) 197 17 9 39.4 (34.5- 44.8) 134 23 10 44.9 (39.5- 50.7) 100 11 11 53.3 (47.1- 59.8) 65 13 12 61.9 (54.6- 69.1) 31 9 13 69.8 (61.2- 78.0) 8 5
147 Table 7.8 summarises the effect of covariates examined on the hazard of death following HIV seroconversion. In univariate analysis, age was the only covariate found to be significantly associated with progression to death.
To examine the effect of all covariates further, models were fitted for each covariate of interest whilst adjusting for the effect of: age at seroconversion, sex and exposure category.
7.5.1 Exposure category
After adjusting for the possible effects of age, male IDUs were found to have a slightly increased non-significant risk of dying compared to men who have sex with men (RR= 1.31, 95% CI= 0.89- 1.93) as did men exposed through sex between men and women (RR= 1.21, 95% CI= 0.53- 2.77). After adjusting for the possible effect of age, women exposed sexually had a significantly lower risk than women exposed parenterally (RR= 0.31,95% CI= 0.11-0.88).
7.5.2 Age
Age was found to be significantly independently associated with progression to death. After adjusting for exposure category and sex, a relative risk of 1.04 per one year increase in age was estimated (95% CI= 1.03- 1.06). This is equivalent to a 51% increase in risk for a 10 year increase in age (95% CI= 29- 78%).
There was no evidence that the risk of death was different between persons aged 16- 24 years and persons aged 25- 34 years. Risk appeared highest in the first 7 years following seroconversion for persons age 25- 34 years. The relative risk of death was estimated to be 0.96 (95% CI= 0.70- 1.34), 1.45 (95% CI= 0.96- 2.19), and 3.83 (95% CI= 2.33- 6.31) respectively for persons aged 16-24, 35- 44, and 45 years or over compared to persons aged 25- 34 years. Five years following seroconversion, progression to death from any cause was estimated to be 11.8%, 7.2%, 13.3%, and 35.1% respectively for age groups 16- 24, 25- 34, 35- 44, and 45 years or over (Figure 7.8 & Table 7.9).
148 Table 7.8 Factors associated with progression from HIV seroconversion to death
Variable Relative Risk (95% Cl)
Univariate Multivariate*
Exposure Sex between men 1.00 1.00 category Sex between men & women 0.67 (0.36,1.25) 0.80 (0.40, 1.58) Injecting drug users 1.10(0.79,1.54) 1.45(1.01,2.10) Others 1.75 (0.64, 4.75) p=0.32 1.81 (0.64, 5.07) p=0.10
Sex Men 1.00 1.00 Women 0.79 (0.53, 1.16) p=0.21 0.85 (0.54, 1.33) p-0.48
Age For each year 1.04(1.02, 1.06) p=0.0001 1.04(1.03, 1.06) p=0.0001
How identified Clinic 1.00 1.00 Laboratory 1.24 (0.90,1.70) p=0.19 1.31 (0.94, 1.82) p=0.12
HIV test interval same calendar mth 1.00 1.00 1-36 mths 0.68 (0.38, 1.22) p=0.22 0.86(0.48, 1.56) p=0.63
Calendar year of For each 0.99 (0.93,1.05) p=0.65 0.99 (0.92, 1.06) p=0.79 seroconversion subsequent year
* Adjusted for exposure category, sex, and age at seroconversion
VO Figure 7.8
Kaplan-Meier plots for time from HIV seroconversion to death from any cause
by age group at HIV seroconversion
1.00
25- 34 yrs
16- 24 yrs
X! X!S &2 .50 -
45 yrs or over 3 35- 44 yrs
.00
1 34 5 6 7 8 9 10 11 12 13
Time from seroconversion (> rs)
Table 7.9
Examining the effect of age at seroconversion on survival estimates Age group (yrs) RR Cumulative proportion dead 5
(95% Cl)* years following seroconversion
(95% Cl)
16- 24 0.96 (0.70- 1.34) 11.8 (7.9-17.4)
25-34 1 7.2 (4.7-10.9)
35- 44 1.45 (0.96-2.19) 13.3 (7.7-22.5)
45 or over 3 83 (2.33-6.31) 35.1 (21.6-53.5)
Log likelihood 24.79,/? « 0.0001 (not adjusted for other factors)
150 A strong age association was also found among men who have sex with men. Compared to those aged 25- 34 years. Relative risks (and 95% confidence intervals) were: 1.13 (0.70- 1.34), 1.97 (1.20- 3.23), and 4.21 (2.39- 7.41) respectively for men aged 16- 24 years, 35- 44 years, and 45 years or over.
7.5.3 Width of HIV test interval
Restriction of analysis to seroconverters with narrower HIV test intervals appeared to have little effect on survival estimates overall and within the first 3 years of seroconversion (Table 7.9 Figure 7.8).
151 Figure 7.9
Kaplan-Meier plots of the effect of the HIV test interval on progression estimates from HIV seroconversion to death
.00
.75
.e ÆîS 2 s. It .50
.25 maximum interval -12 months
maximum interval- 24 months
maximum interval -36 months
.00
1 2 3 45 6 7 8 9 10 11 12 13 14
Time from seroconversion (yrs)
Table 7.10 The effect of narrower HIV test intervals on HIV survival estimates within the first 3 years of HIV seroconversion
Time from Cumulative progression estimates to death (95% Cl) seroconversion Seroconverters with Seroconverters with Seroconverters with (yn) 3- yr maximum 2- yr maximum 1- yr maximum interval interval interval Î 09(0.4-2.1) 1.1 (0.5-2.4) 1.5 (0.7-3.3) 2 2.6 (1.6- 4.1) 2.6(1 6-4.3) 2.4 (1.3-4.5) 3 3.5 (2.3- 5.2) 3.5 (2.3- 5.4) 3.8(23- 6.2)
152 CHAPTER VIII
A SUMMARY OF FINDINGS AND FUTURE WORK
This chapter summarises findings reported in this thesis, places them in the context of work done by other investigators, and outlines work yet to be done outside its scope.
The major advantages of this work are that it has resulted in the availability of a large multi-centre, multi-risk group national cohort, with a wide age range and with a low lost to follow up rate. Because most subjects were ascertained retrospectively, it was possible to assemble it in a fairly short space of time. Further, the prospective ascertainment of seroconverters will enable continual and unbiased estimates to be made of the incubation period distribution. Methodological aspects have been and are continually assessed so that sources of possible bias are highlighted and dealt with in ensuing analyses.
A number of studies of HIV seroconverters have published findings during the time this work was being undertaken. As well as providing updates of progression estimates derived from various cohorts of seroconverters, a number of authors have drawn attention to issues of methodology. Findings from these are also included here.
8.1 Overall estimates
The probabilities of progression to AIDS and death found here are in general agreement with those from other studies, particularly those of similar population structure, i.e. predominantly homosexual and bisexual men (Figure 8.1).
Estimates of the AIDS incubation period and of the time from seroconversion to death are, however, meaningless unless they are provided together with the age distribution of the population to which these estimates relate.
153 Figure 8.1
Cumulative proportion of persons developing AIDS by time since seroconversion; estimates from the Register compared to estimates from other cohorts
70
60
s 5 S 40 c I
20 The UK Register
Other cohorts
4 6 8 10 12 14
Time from seroconversion (yrs)
154 The possible bias resulting from the preferential inclusion of recent seroconverters and of long-term survivors bas been dealt with as much as possible. The rate of loss to follow up bas also been reduced as this often differs between exposure categories and can lead to misleading estimates between injecting drug users and homosexual men, for example.
8.2 The effects of exposure category and sex
No evidence of a difference in AIDS progression estimates by exposure category or sex were found in the Register.
After adjusting for the effects of age and sex, the lower relative risk of progressing to AIDS for IDUs compared with men who have sex with men was found to be statistically non-significant. Most studies have reported similar findings Table 8.1 summarises the relative hazard of progressing to AIDS from published studies comparing these two exposure categories.
This study found no evidence of a difference between persons exposed through sex between men and those exposed through sex between men and women after adjusting for age and sex (RR= 0.77, 95% CI= 0.37- 1.60). Investigators of the SEROCO study in France, however, reported a higher hazard for homosexual men compared to heterosexuals after adjusting for age and sex (RR= 2.42, 95% CI= 1.2- 4.9) Their finding may have been due to some extent to the high proportion of KS diagnosed in the homosexual men (40%- 19 of 47).
8.3 The effect of age
Faster progression to AIDS and death with increasing age at seroconversion has been observed in a number of published studies and gives credence to the likely importance of age as a cofactor for progression (Chapter II). Most studies of homosexual and bisexual men, however, have not observed this effect due to a lack of heterogeneity in the age structure of the seroconverters in those studies. The increase in risk of AIDS in
155 Table 8.1 The relative risk of AIDS for injecting drug users compared to men who have sex with men: findings from published studies
Study RR for IDUs compared with men who
have sex with m en| (95% Cl)
UK Register 0.74 (0.48- 1.12) European Seroconverter Study & 0.83 (0.60- 1.15) Tricontinental Seroconverter Study Oslo HIV Cohort study 0.50(0.14- 1.67) Italian Seroconversion Study (ISS) 1.02 (0.71- 1.45) j adjusted for age and sex
156 the present study is comparable to the findings from other cohort studies (Table 8.2). Data from the Register gives an estimate of a 43% increase in the risk of AIDS for each 10-year increase in age at seroconversion. The increase in risk of progression to death for each 10-year increase in age is estimated to be 1.51 (95% CI= 1.29- 1.78). This relative risk is similar to findings reported by Hessol et al of 1.65 (95% CI= 1.12- 2.41) and Veugelers et al of 1.50 (1.06- 2.11).
8.4 The effect of right censoring strategies
One of the major factors contributing to differences in estimates of the incubation period between studies and exposure categories appears to be the choice of right censoring strategy. This was illustrated in Chapter VI by using a number of different strategies on data from the Register. This led to estimates for the median time to AIDS of between 8.1 years and 10.8 years depending on which strategy was used. It is, therefore, important that the choice of strategy is made a priori.
If it is assumed that all persons lost to follow up, even if for many years, are still alive and/or AIDS-free at the time of analysis, this may lead to an over-estimate of the length of time from seroconversion to death and/or AIDS. Conversely, if each member of the cohort is censored at the date of their last clinic visit persons with better prognosis may be preferentially excluded if they are unlikely to attend as frequently as persons who were ill at their last clinic visit. This would lead to under-estimating the length of time from seroconversion to death and/or AIDS.
In this thesis I proposed a strategy to overcome these assumptions and minimise the bias resulting from making them. As the AIDS surveillance registries in the UK are estimated to be 87% complete, and as 95% of AIDS cases are reported within 2 years of diagnosis, it is valid to assume, after allowing a 2-year reporting delay, that persons not on these registries are AIDS-free. For HIV survival estimates, cross checks were made against the national death register at ONS. As this is 100% complete, it may be reasonable to assume that persons not on the ONS file are alive. This assumption, however, assumes that no-one died outside the UK and that the person identifiers used in the matching process were the same as those given by the informant who reported the death.
157 Table 8.2 The relative risk of AIDS of a 10-year increase in age at seroconversion from published studies
Author Relative risk Exposure category(ies)
(95% confidence intervals) included
UK Register 1.43 (1.19,1.71) Homosexual, heterosexual. IDU Phillips 1.45 (1.15,1.85) Haemophiliac Hessol 1.31 (0.95, 1.8) Homosexual Veugelers 1.11 (0.85, 1.46) Homosexual Rosenberg 1.61 (1.26-2.07) Haemophiliac Rosenberg 1.36(0.90- 2.04) Homosexual Carré 1.34(1.03, 1.77) Homosexual, heterosexual Pezzotti 1.55 (1.32, 1.83) Homosexual, IDU Operskalski 1.34 Homosexual, Haemophiliac, Transfused Munoz 1.14 Homosexual
158 Two other studies proposed alternative right censoring strategies. Brettleet al reported on data on 260 IDUs with narrow seroconversion intervals from the Edinburgh City Hospital cohort. AIDS-free participants were differentially censored according to whether they were ‘active’ (last seen within the last year of the study- 24th March 1994), or ‘inactive’ (not seen for over one year). Active participants were censored at 24 March 1994 and those inactive censored at the last clinic visit. For survival analysis, a similar censoring strategy was employed with active participants censored at 24 March 1994 and inactive participants censored at 6 months earlier (to allow a six-month lag in death reports from the Registrar General- the central registry for births and deaths in Scotland).
A group in the Netherlands examined the effect of right censoring strategy on progression estimates using data on 422 men registered in the Tricontinental Study
Depending on which right-censoring strategy was applied, 3 different estimates of the median time from seroconversion to the development of AIDS were produced: 8.5, 9.2 and 10.2 years. Under Strategy 1 persons were censored as AIDS-free at the date of their last visit, producing the shortest median time of 8.5 years (95% CI= 7.6- 9.6) was derived. Under Strategy 2 persons lost to follow up were censored as AIDS-free at the date of close of analysis file (1 January 1995) and a median of 10.2 years (95% CI= 9.0- oo) was derived. Under Strategy 3, the authors’ proposed and preferred option, persons not seen in the year preceding 1 January 1995 as well as those with an AIDS diagnosis greater than one year after their last visit (as ascertained through linkage with AIDS registries) were censored as AIDS-free one year after the last visit. Persons seen within one year of 1 January 1995 and remaining AIDS-free are censored at the date of the last visit and persons developing AIDS within one year of the last visit are censored at the date of AIDS diagnosis. This third strategy produced an estimate of a median time to AIDS from seroconversion of 9.2 years (95% CI= 8.4- oo).
A number of other researchers agree that right censoring strategies are important and that the classical approach of censoring persons who are event-free at the time they were last seen may not be valid and may result in misleading estimates.
159 8.5 The effect of late entry
Similar to a number of other cohorts of seroconverters, most subjects included in the Register have been identified retrospectively. As this may potentially mean that longer- term survivors may have been preferentially followed up, there is a danger that an analysis approach which ignores this will under-estimate progression estimates. The size of this under-estimate depends on the rate of death in those not included compared to those who are. Careful documentation of a date of entry into unbiased follow up for each seroconverter has enabled correction for this potentially serious source of bias. Although such an approach to analysis overcomes this bias, it resulted in marginally reduced certainty in the estimates of progression rates within the first few years following seroconversion as the number of subjects followed up from seroconversion is fewer when the strategy of delayed entry is used.
An analysis which allows for late entry requires a great deal of effort to establish when each individual should enter the risk set but in this case it resulted in marginally different results fi*om estimates which ignore it. This does not, however, mean that the issue of left censoring can safely be ignored.
8.6 Temporal trends in the incubation period distribution and HIV survival
There was little evidence from the data reported in this work of a change over calendar time in the incubation period distribution or in survival from HIV seroconversion. As the management of infected individuals advances, it is hoped that the time from seroconversion to both AIDS and death will become longer. The advantage of the Register is that it will continue to recruit new seroconverters and to follow up those already recruited. It will, therefore, provide an opportunity to continually evaluate any changes over time. As with other studies of seroconverters, however, it is likely to be underpowered to evaluate changes in the most recent time periods as few events and person years at risk will be included in these periods.
It is recognised that the changes to the AIDS case definition, which allow AIDS to be
160 diagnosed at an earlier stage, may result in an apparent shortening in the IPD (i.e. lower estimates in the more recent time periods). Any such changes in the time from seroconversion to death, however, are not affected by this.
A number of studies have reported an increase in the hazard of developing AIDS over time. Hessol et al reported a significant increase in hazard for persons seroconverting between July 1989 to December 1991 compared to those who seroconverted before 1987 (RR= 1.77, 95% CI= 1.01- 3.10) Veugelers et al reported a non-significant increase in the hazard over time to 1986 They further reported on an increase in the hazard to death for persons who seroconverted in 1985 relative to those who seroconverted before 1984 (RR= 2.36, 95% CI= 1.00- 5.54).
More recently, a group from Turin, Italy, one of the clinical centres participating in the ISS, reported a significant increase in progression rates to a CD4 cell count of 500, 400 and 200 per ml, as well as the diagnosis of AIDS in those most recently infected Their findings appeared to be corroborated by more rapid decline in CD4 cell counts in persons who seroconverted in the more recent time periods. This led the authors to conclude that this was evidence of the emergence of more virulent strains of the virus responsible for more aggressive course of HIV disease in patients recently seroconverted. This finding may, however, be due to the fact that persons who presented with high CD4 counts were more likely to be excluded from the analysis if they first presented in the more recent time periods.
Other studies have reported a decrease in the hazard of progressing to AIDS over calendar time. Taylor et al^^ reported a statistically non-significant decrease from the first half of 1979 to the first half of 1983. Muhoz and Xu reported that for persons with a CD4 cell count of 100 cells/ ml or less the hazard to the development of AIDS had decreased over calendar period with relative hazards of 0.66 (95% CI= 0.51- 0.85) and 0.60 (95% CI= 0.47- 0.79) for those observed in 1988- 1991 and after 1991 respectively compared with those observed prior to 1988
An improvement in progression rates over calendar time was also reported by the
161 investigators of the Amsterdam cohort of homosexual men The authors analysed data on 39 men who seroconverted in 1984 and 1985 (early) with 42 men who seroconverted between 1986 and 1989 (middle) and 34 men who seroconverted between 1989 and 1993 (recent). The 5-year AJDS-ffee survival was found to be 55% in the early seroconverters and 77% in the middle group (p= 0.09). Five year survival was 81% in the early and 97% in the middle groups (p= 0.04) suggesting an elongation of the incubation period and of the time from seroconversion to death for persons who seroconverted between 1985 and 1993.
All these studies provide conflicting accounts of changes in the hazard of AIDS and death over time. Findings may be due to preferential over-ascertainment of events (for those who reported a decrease in hazard over time) or under-ascertainment (for those who reported an increase in hazard over time) in the earlier time periods. Further, if the rate of loss to follow up is outcome dependent (with those lost to follow up having better prognosis- Section 4.6.1), and this is greater for those who seroconverted in the earlier time periods, this will lead to an observation of a change in the hazard over time. If persons lost to follow up are censored at the date they were last seen, a decrease will be observed. If they are assumed to be event-free at the analysis cut-off date, an increase will be observed. Findings may also be due to chance and the likelihood of them being real are strengthened if they are repeated in other settings. For the calendar period 1983 to 1995 most studies of seroconverters have reported no strong evidence of a temporal change in the incubation period or of survival.
8.7 Further work
As persons recruited to the Register now are prospectively ascertained, this will lead to more reliable progression estimates of the first few years following seroconversion. And, as the follow up of those recruited continues, the rate beyond 10 years from seroconversion can be estimated with greater certainty. A fall in the number of subjects recruited has been observed from 1994 (Figure 5.3) which may, in part, be due to a reporting lag. It may also be due to a reduction in HIV incidence, but this is difficult to disentangle from a fall in the numbers of infected persons with a history of a previous negative test. Effort is continuing to ensure the complete ascertainment of all eligible seroconverters.
162 Encouraging results from clinical trials suggest that the use of dual anti-retroviral therapy in the mid-1990s is likely to prolong survival and, depending on uptake rate, this would be expected to lead to an observation of better prognosis in the more recent time periods. Further improvements are anticipated with increasing use of drug regimens which include two nucleoside analogues and a protease inhibitor. As the majority of subjects on therapy included in the analyses presented in this thesis are likely to have received monotherapy, it may be too early for the benefits of combination therapy to be observed in terms of the length of time from seroconversion to AIDS and, more importantly, of improving survival.
Information available on anti-retroviral therapy and PCP prophylaxis for seroconverters included in the analyses presented in this work are incomplete and could not be included in this work. While treatment information is fairly complete for new seroconverters, for persons who seroconverted some time before the Register was set up or who had died, little to no information on what treatments had been prescribed to them was reported by the clinics. During the course of this work all centres were contacted regarding missing treatment information and this is expected to now be complete. The possible effect of anti-retroviral treatment on progression estimates and over time can be assessed by including them in statistical models as time-dependent covariâtes.
CD4 data have been shown to be reliable prognostic markers of HIV disease progression It will be important to monitor the rate of decline in CD4 with any changes observed over time in the clinical progression of those infected. An error in the original Register report forms requested information on the latest two CD4 counts for seroconverters. For persons ascertained retrospectively (at present the majority of seroconverters on the Register) this meant that information on CD4 cell counts close to the time of seroconversion was not provided. This has now been rectified and CD4 cell information will be available on most seroconveters from 1988 onwards.
As a number of outstanding questions may require answers which are age-specific and exposure category spécifié, no one study alone is likely to be able to address them. My work in the future will not only be concerned with upkeeping and examining data from the Register but also in pooling data from it with data from other similar cohorts
163 nationally and internationally. A preliminary analysis of data from the UK Register with data from the UK Haemophilia Centre Directors’ Organisation is underway. The main aim of this analysis is to provide age-specific estimates of the incubation period distribution for use in the work of AIDS projections in England and Wales.
On an international level, a grant from the UK Department of Health has allowed the investigators of the Register, Imperial Cancer Research Fund (ICRF) Cancer Epidemiology Unit in Oxford, and the MRC Biostatsitics Unit in Cambridge to hold a Workshop in April 1997. Data on over 13,000 seroconverters from 35 cohorts in Europe, North America and Australia were pooled and I am currently involved in the analyses together with colleagues from ICRF. I am also responsible for the co ordination of a Concerted Action resulting from a successful bid to the European Union Biomed 2 programme. Its main aim is to pool European data and expertise to monitor the incubation period and survival, and in particular changes in recent time periods.
Methods of recruitment into these cohorts vary, as do follow up procedures and completeness of ascertainment of events (e.g. AIDS and death). It is important to understand and document all these issues, as they present potential sources of bias, so that they can be adjusted for, wherever possible, in the statistical analysis.
In conclusion, a national register of HIV seroconverters has been established in the UK and is now one of the largest cohorts in the world capable of monitoring the times from HIV seroconversion to AIDS and to death and the factors which may influence them. The Register has established a high rate of long-term follow up through matching against a number of national databases of HIV infected persons and of all deaths in the country.
As the recruitment of newly infected individuals continues, essential questions on the evolution of the clinical course of HIV disease can be addressed in a timely way in the future.
164 Appendix I
Members of the Executive and Steering Committees
Executive Committee:
Dr Abdel Babiker- MRC HIV Clinical Trials Centre, UCLMS Professor Janet Darbyshire MRC HIV Clinical Trials Centre, UCLMS Dr Barry G. Evansj- PHLS Communicable Disease Surveillance Centre Dr Noel Gillj:- PHLS Communicable Disease Surveillance Centre Professor Anne Johnson- MRC UK HIV Epidemiology Co-ordinating Centre, UCLMS Professor Andrew Phillips- Royal Free Hospital School of Medicine
Steering Committee:
Professor Valerie Beral (Chair)- ICRF Cancer Epidemiology Unit, Oxford Dr Ray Brettle- City Hospital, Edinburgh Dr Chris Came- Addenbrooke’s Hospital, Cambridge Dr Richard Gilson- Academic Dept of Sexually Transmitted Diseases, UCLMS Dr David Goldberg- Scottish Centre for Infection and Environmental Health, Glasgow Dr David Hawkins- Chelsea & Westminster Hospital, London Professor Don Jeffries- St Bartholomew’s Hospital, London Dr Margaret Johnson- Royal Free Hospital School of Medicine, London Dr Anton Pozniak- King’s College Hospital, London Professor Andrew J. McMichael- Institute of Molecular Medicine, Oxford Dr Philip Mortimer- PHLS Central Public Health Laboratory, London Professor Jonathan Weber- St Mary’s Hospital, London Ms Sally Wellsteed- UK Department of Health, London
t until 1996 j from 1996
165 Appendix II
Names of collaborators
The following is a list of collaborators who have helped identify seroconverters and who have provided information on them.
Aberdeen: City Hospital- Dr C C Smith, Dr RBS Laing Airdrie: Monklands Hospital- Dr WTA Todd Ashford: Ashford Hospital- Dr N Kirk Barnsley: District General Hospital- Dr J Dhar, SC Allen Basingstoke: District Hospital- Dr DH Jackson Belfast: Royal Victoria Hospital- Maeve O'Gorman, Dr R Maw, Sinead McKeman, Connal McCaughey Birmingham: Heartlands Hospital - Gill Cooper, Dr C Ellis, Sarah Hackforth, Dr S Skidmore General Hospital- Gerry Gilleran, Dr M Shahmanesh Blackpool: Victoria Hospital- Michelle Jenkins, Dr A Saeed Bolton: General Hospital- Dr E Curless, Dr D Nunns Bournemouth: General Hospital- Dr R Basu Roy, Dr MJ Hayward, Jacqui Campbell Brighton: Claude Nicol Centre- Dr F Clark, Dr E O’Moore, Julie McIntosh, Dr S Tchamouroff, Dr B Thom. Bristol: Royal Infirmary- Dr P Homer Southmead Hospital- Dr S Glover, Dr A Roome Cambridge: Addenbrooke's Hospital- Dr C Came, Dr T Wreghitt Chertsey: St. Peters Hospital- DR JM Pritchard, Dr C Grundy Chester: City Hospital- Dr C O’Mahoney Chesterfield: Chesterfield & N Derbyshire Royal Hospital- Dr K Rogstad Chichester: St. Richards Hospital- Dr M Greig, Barbara Hayman, Dr T Bushell Colchester: Essex County Hospital- Dr Thavabalan Doncaster: Royal Infirmary- Dr Moss, Dr Nathan Dundee: Royal Infirmary- Dr A France Durham: Drybum Hospital- Dr C White Edinburgh: City Hospital- Dr R Brettle Royal Infirmary- Dr G Scott, Dr S McMillan Glasgow: Ruchill Hospital- Dr D Kennedy, Dr A Pithie, Cathy Graeme Royal Infirmary- Dr A Scoular Southem General- Dr JM Harvey Gloucester: Royal Hospital- Dr Butland, Barbara Watson, Dr Suleiman Greenwich: District Hospital- Dr S Mitchell, Dr J Russell Guildford: Famham Road Hospital- Dr L Howard High Wycombe: General Hospital- Dr GA Luzzi Huddersfield: Royal Infirmary- Dr SS Al-Egaily Isleworth: W Middlesex Hospital- Dr M Anderson Kilmamock: Crosshouse Hospital- Dr G Williams Leeds: General Infirmary- Dr C Lacey, Kathryn Short, Dr C Newman, Gerald Booth Leicester: Royal Infirmary- Dr P Schober London: Ambrose King Centre- Samantha Flint 166 Archway Clinic- Dr P Kell, Rose Tobin Central Middlesex Hospital- Dr G Brook, Marion Bond Charing Cross Hospital- Dr K McLean Ealing Hospital- Dr W Lynn FACTS Medical Centre- Dr K Smith Hammersmith Hospital- Dr I Ushiro, Dr S Shaunak Kings College Hospital- Dee Graham Kohler Centre- Ramesh Halai, Leslie Simpson Lewisham Hospital- Dr Rao Mortimer Market Centre- Julie Dodds, Tony Nardone, David Comforth, Dr DWilliams, Ruth Johnstone Newham General- Dr J Del Amo, Dr A Wisdom North Middlesex Hospital- Dr Y Drabu Royal Free Hospital- Dr Amanda Mocroft, Dr M Johnson St. George’s Hospital- Dr F Davidson, Jill Walker St. Mary's Hospital- Els Vermeulen, Katy Wood, Gail Birse-Said, Dr D Churchill St. Thomas’s Hospital- Dr B Peters, Dr C Bradbeer Victoria Sexual Health Clinic- Dr S McCormack West Middlesex Hospital- Dr Mahenra Whipps Cross Hospital- Dr R Melville Whittington Hospital- Dr N Parker, Dr M Kelsey Luton: Luton and Dunstable Hospital- Dr T Blachandran Manchester: Hope Hospital- Dr B Goomey North Manchester General Hospital- Cynthia Murphy Withington Hospital- Dr S Chandiok, Maureen Lamb, Dr A Turner Manchester Royal Infirmary- Dr B Mandai Melrose: Borders General Hospital- Dr MI Brown Middlesborough: Middlesborough Hospital- Dr A Opaneye Milton Keynes: Milton Keynes Hospital- Dr J Haynes Newcastle: General Hospital- Joan Wilson, Dr E Ong, Dr N Lightfoot, Dr P Watson Newport: Royal Gwent Hospital- Dr R Das Northampton: General Hospital- Dr A Kilvert Norwich Hospital- Margaret Sillis, Dr J Meaden Nottingham: City Hospital- Dr CJ Bignell Oldham: Royal Oldham Hospital- Dr Girgis Plymouth: General Hospital- Dr JR Willcox, Dr D Dance Portsmouth: St. Mary’s Hospital- Dr JM Tobin, Lynda Tucker Preston: Royal Hospital- Dr Morgan-Capner Reading: Royal Berkshire Hospital- Dr A Tang Redhill: New East Surrey Hospital- Dr Ramanathan Rochdale: Baillie Street Health Centre- Dr H Lacey Sheffield: Royal Hallamshire- Dr G Kinghom, Sandra Herman, Dr M Schmid Northern General Hospital- Dr G Kudesia Slough: Upton Park Hospital- Dr S Dawson Southampton: Royal South Hampshire Hospital- Dr F Wilmott, Dr S O’Connell Stoke-on-Trent: North Staffordshire Hospital- Dr S Sivapalan Sutton-in-Ashfield: King’s Mill Centre- Dr M Rahman 167 Swansea: Singleton Hospital- Dr K Yoganathan Swindon: Princess Margaret Hospital- Dr C Rabindran Telford: Princess Royal Hospital- Dr S Devendra Thornton Heath: Mayday Hospital- Dr. P Deheragoda, Dr Sahathevan Torquay: Torbay Hospital- Dr IG McGill Walsall: Manor Hospital- Dr AT Joseph Watford General Hospital- Dr P Munday Wigan: Royal Albert Edward Infirmary- Dr J Forrer Wolverhampton: New Cross Hospital- Dr S Sivakumaran Worcester: Royal Infirmary- Dr Crooks
168 Appendix III
Definition of ’acute infection’ a) An individual believed to be at increased risk for HIV infection presenting with an illness compatible with acute primary HIV infection without documentation of a negative HTV antibody test will be enrolled into the study pending the outcome of laboratory investigations to establish evidence of recently acquired infection, as detailed below. When the results of the investigations are available, individuals not meeting these criteria will be removed from the Register. b) Any other HTV positive individual without a previous negative test in whom the laboratory criteria below have been met will be enrolled.
Laboratory criteria of recent seroconversion (one criterion required): a) The presence of p24 antigen, confirmed by neutralisation, and/or PCR positivity, in the absence of total anti-HIV at that time, but in whom anti-HIV is demonstrated in a specimen collected within 6 months. b) Weak reactivity in EIA or low titre (<1/256) by Serodia-HIV and IgM anti-HIV or p24 antigen and seroconversion-type Western Blot pattern, with either: increasing titres of anti-HIV/EIA reactivity in specimens collected within a 3-month period but tested simultaneously, or development of new bands or intensification of weakly reactive bands in 2 or more sequential sera collected at an interval of no more than 3 months and tested contemporaneously by Western Blot.
169 Appendix IV
Form 1 UK REGISTER OF HIV SEROCONVERTERS INITIAL REGISTRATION FORM
Clinic no.: Date of birth: îSèroCTttverter^^^" ^ ^ Register no.-
Full initials: Soundex code: Sex:
Ethnic group (tick one only): Probable route of transmission (tick one or more): □ White □ Homosexual/bisexual sex □ Black African □ Injecting Drug Use (IDU) □ Black Caribbean □ Heterosexual sex □ Indian/Pakistani/Bangladeshi □ Other (specify)...... □ Other (Specify)......
What is the likely country of infection?
Was a seroconversion-type illness documented? □ Yes □ No
If Yes, Date ______Main symptoms ______
Date last negative HTV antibody test 6. Date first positive HTV antibody test
Name of hospital or lab Name of hospital or lab
Is patient considered likely to be in the acute stage of HTV infection? □ Yes □ No If Yes, give details mcluding laboratory evidence such as the presence of p24 antigen m the absence of HIV antibody, etc.
Results of the two most recent CD4 cell counts Count 107ml Date _ Count lO^'ml Date_ Results of two most recent plasma viral load measurements Count copies/ml Date__ Count copies/ml Date__
10 . Has antiretroviral therapy ever been prescribed? □ Yes □ No
Drug Date first started Date patient was last on drug Zidovudine (AZT)
Didanosine (ddD
Dideoxycytidine (ddC)
Other (specify)
11. Has any prophylaxis or maintenance therapy for opportunistic infections ever been presribed? □ Yes □ No
Drug Date first started Date patient was last on drug
170 Form 1 12. Have any of the following AIDS defining conditions been diagnosed □ Yes □ No If Yes, indicate for each disease the date and method of diagnosis (Definitive or Presumptive)
Diseases Diagnosis Diseases Diagnosis Date Method Date Method OPPORTUNISTIC INFECTIONS tuberculosis (extrapulmonaiy) D P
Candidiasis: / D other (disseminated) D P trachea/bronchi/Iungs oesophageal D P Pneumocystis carinil pneumonia D P Cerebral toxoplasmosis D P Pneumonia D P (2 episodes in 12-mth period) Cocddlodomycosis / D Prog multifocal leukocncephalopatby / D (extiapulnxmaiy) Cryptococcosis / D Salmonella septicaemia (recurrent) D (extrapulmonaiy) Cryptosporldlosls / D (duration > I mth) MALIGNANCIES
Cytomegalovims Cervical cardnoma (invasive) . / D retinitis / D P other / D Kaposi's sarcoma D P HSV (duration > I mth) / D NotHHodgldn's lymphoma primary in brain / D Histoplasmosis D BurtdtTs or equivalent term / D (disseminated or extrapulmonaiy) Isosporlasis diarrhoea / D immunoblastic or equivalent / D (duration > I mth) Mycobacterium avium (extrapulmonaiy) D P OTHER tuberculosis (pulmonary) / D P AIDS demenda Complex / D HTV Wasting syndrome / D
13. Date patient last assessed in the clinic
14. Has the patient died? □ Yes □ No
If Yes, Date of death ______Certified cause
Do you consider the death to be (tick one) and give details below:
□ HTV related □ Drug related □ Possibly HTV and/or drug related
□ Not HTV or drug related □ Suicide □ Uncertain
Please give details of terminal illness
15. Blood specimens
Has a 10 ml specimen been taken for the Register for serum storage locally?
□ Yes Date □ No
Signed: Centre: Today’s date:
(Print name); Consultant;
Please send this form in a pre-addressed envelope to: Dr Janet H Darbyshire, MRC HIV Clinical Trials Centre, UCLMS, The Mortimer Market Centre, Mortimer Market (off Capper St), London WCIE 6AU. Tel 0171-380 9991/67. Fax 0171-380 9972. 171 Form 2a UK REGISTER OF HIV SEROCONVERTERS ANNUAL FOLLOW UP FORM
(For seroconverters not yet on the Register please use the initial registration form- Form 1)
Clinic no.: Date of birth: Seroconverter register no.: Full initials: Soundex code: Sex:
1. Date patient last assessed in the clinic
Results of the two most recent CD4 cell counts Count 107ml Date
Count 107ml Date
3. Results of two most recent plasma viral load measurements Count ______copies/ml Date
Count copies/ml Date
Please give detail of anti-retroviral therapy prescribed in last 12 months
Tick here if none □
Drug Date first started Date patient was last on drug Zidovudine (AZT) Didanosine (ddl) Dideoxycytidine (ddC) Other (specify)
5. Please give detail of any prophylaxis or maintenance therapy for opportunistic infections presribed in last 12 months
Tick here if none □
Drug Date first started Date patient was last on drug
172 Form 2a 6. Have any of the following AIDS defining conditions been diagnosed □ Yes □ No If Yes, indicate for each disease the date and method o f diagnosis (Definitive or Presumptive)
Diseases Diagnosis Diseases Diagnosis
Date Method Date Method OPPORTUNISTIC INFECTIONS tuberculosis (extrapulmonaiy) D P
Candidiasis: / D other (disseminated) / D P tiacheaAsronchi/lungs oesophageal D P Pneumocystis carinil pneumonia D P Cerebral toxoplasmosis / D P Pneumonia / D P (2 episodes in 12-mth period) Cocddlodomycosis D Prog muidfocal leukoencephaiopathy D (extrapuimonary) Cryptococcosis D Salmonella sepdcaemia (reçurent) / D (extrapulmonaiy) Cryptosporidlosis D (duration > 1 mth) MALIGNANCIES
Cytomegaiovirus Cervical carcinoma (invasive) / D letinitis / D P other / p Kaposi's sarcoma J D P HSV (duration > 1 mth) / D Non-Hodgkin's lymphoma primary in brain / D Histoplasmosis D Buridtfs or equivalent term / D (disseminated or extrapulmonaiy) Isosporlasis diarrhoea / D immunoblastic or equivalent D (duration > 1 mth) Mycobacterium avium (extrapulmonaiy) / D P OTHER tuberculosis (pulmonary) D P AIDS dementia Complex / D HTV Wasting syndrome / D
7. Has the patient died? □ Yes □ No
If Yes, Date of death Certified cause
Do you consider the death to be (tick one) and give details below:
□ HIV related □ Dmg related □ Possibly HTV and/or drug related
□ Not HTV or drug related □ Suicide □ Uncertain
Please give details o f terminal illness
Blood specimens
Has a further 10 ml specimen been taken for the Register for serum storage locally?
□ Yes Date □ No
Signed: Centre: Today’s date:
(Print name): Consultant:
Please send this form in a pre-addressed envelope to; Dr Janet H Darbyshire, MRC HTV Clinical Trials Centre, UCLMS, The Mortimer Market Centre, Mortimer Market (off Capper St), London WCIE 6AU. Tel 0171-380 9991/67. Fax 0171-380 9972.
173 Appendix V
Time of entry into risk set for each clinical centre
For each clinical centre a review was made of the completeness of inclusion of all eligible seroconverters and a new date field ‘enrolled’ was imputed. This field signifies the date at which seroconverters from that centre enter the risk set in the analysis of times from HIV seroconversion to AIDS and to death (see Chapter IV- Left truncation of survival times from seroconversion). For centres with complete ascertainment, date ‘enrolled’ is the date of seroconversion, and for centres where ascertainment is incomplete, the ‘enrolled’ date is the date from the time point at which it is believed to be complete. If ascertainment from a centre is not complete from any point in time, or is uncertain, follow up for all seroconverters from that centre commences from the date of registration.
Clinical centre Extent to which all eligible seroconverters have been identified and included in the Register Mayday Hospital Random recall and/or list of those identified through laboratory data at CDSC. Date enrolled = date registered Archway Clinic Random recall and/or list of those identified through laboratory data at CDSC. Date enrolled = date registered Bristol Royal Infirmary Random recall and/or list of those identified through laboratory data at CDSC. Date enrolled = date registered Hope Hospital Through comprehensive search by Manchester PHL of all seroconverters identified through Manchester and Preston PHLs and North Manchester Regional Virus Laboratory Date enrolled = seroconversion date North Manchester Through comprehensive search by Manchester PHL of all General Hospital seroconverters identified through Manchester and Preston PHLs and North Manchester Regional Virus Laboratory Date enrolled = seroconversion date Withington Hospital Through comprehensive search by Manchester PHL of all seroconverters identified through Manchester and Preston PHLs and North Manchester Regional Virus Laboratory Date enrolled = seroconversion date Dundee Royal Infirmary Identified through Scottish Denominator study of all seroconverters in Scotland. Date enrolled = seroconversion date Heartlands Hospital Complete cohort as all records ever have been searched to identify all those eligible. Date enrolled = seroconversion date Royal Victoria Hospital, Only those attending from 27th July 1995 included. Belfast Date enrolled = 27th July 1995 ______
174 Clinical centre Extent to which all eligible seroconverters have been identified and included in the Register______Claude Nicol Centre Only those attending from 27th June 1995 included. Date enrolled = 27th June 1995 Norwich Hospital All seroconverters identified through England & Wales Denominator Study. Not all have been reported, however. Date enrolled = date registered Nottingham City Random recall and/or list of those identified through Hospital laboratory data at CDSC. Date enrolled = date registered Addenbrooke’s Hospital All seroconverters identified through England & Wales Denominator Study. Not all have been reported, however. Date enrolled = date registered Edinburgh City Hospital Edinburgh City Hospital Clinic Cohort- stored sera were tested when an attender presented at the clinic for HIV services. Date enrolled = date of first clinic visit Edinburgh Royal Identified through Scottish Denominator study of all Infirmary seroconverters in Scotland. Date enrolled = seroconversion date Ruchill Hospital Identified through Scottish Denominator study of all seroconverters in Scotland. Date enrolled = seroconversion date Leeds General Infirmary Complete cohort as all records ever have been searched to identify all those eligible. Date enrolled = seroconversion date Newham Hospital Complete cohort as all records ever have been searched to identify all those eligible. Date enrolled = seroconversion date Newcastle General Complete search done only on those who had died, those Hospital still attending, however are randomly reported, are patchy at the moment. Date enrolled = date registered Bolton General Hospital Through comprehensive search by Manchester PHL of all seroconverters identified through Manchester and Preston PHLs and North Manchester Regional Virus Laboratory Date enrolled = seroconversion date Royal Hallamshire Complete cohort as all records ever have been searched to Hospital identify all those eligible. Date enrolled = seroconversion date Southmead Hospital Only those attending from 6th July 1995 included. Date enrolled = 6th July 1995 St Mary’s Hospital, Only current attenders identified and registered Portsmouth Date enrolled = date registered Mortimer Market Centre Only those attending from 1st February 1993 included. Date enrolled = 1st February 1993 ______
175 Clinical centre Extent to which all eligible seroconverters have been identified and included in the Register______Ambrose King Centre Only current attenders identified and registered Date enrolled = date registered Hammersmith Hospital Random recall and/or list of those identified through laboratory data at CDSC. Date enrolled = date registered St Mary’s Hospital, Only those attending from 1st September 1991 included. London Date enrolled = 1st September 1991 King’s College Hospital All seroconverters identified through England & Wales Denominator Study and all have been reported Date enrolled = seroconversion date Kohler Centre Only those attending from 1st January 1990 included. Date enrolled = 1st January 1990 St Thomas’s Hospital Random recall and/or list of those identified through laboratory data at CDSC. Date enrolled = date registered St George’s Hospital Random recall and/or list of those identified through laboratory data at CDSC. Date enrolled = date registered Victoria Sexual Health Only those attending from 1st January 1990 included. Clinic Date enrolled = 1st January 1990 Royal Free Hospital Random recall and/or list of those identified through laboratory data at CDSC. Date enrolled = date registered Bournemouth General Only current attenders identified and registered Hospital Date enrolled = date registered Charing Cross Hospital Complete cohort as all records ever have been searched to identify all those eligible. Date enrolled = seroconversion date Leicester Royal Random recall and/or list of those identified through Infirmary laboratory data at CDSC. Date enrolled = date registered Doncaster Royal Only current attenders identified and registered Infirmary Date enrolled = date registered FACTS Medical Centre Random recall and/or list of those identified through laboratory data at CDSC. Date enrolled = date registered Ealing Hospital Random recall and/or list of those identified through laboratory data at CDSC. Date enrolled = date registered Torbay Hospital Random recall and/or list of those identified through laboratory data at CDSC. Date enrolled = date registered ______
176 Clinical centre Extent to which all eligible seroconverters have been identified and included in the Register______Royal Albert Edward Through comprehensive search by Manchester PHL of all Infirmary seroconverters identified through Manchester and Preston PHLs and North Manchester Regional Virus Laboratory Date enrolled = seroconversion date Birmingham General Random recall and/or list of those identified through Hospital laboratory data at CDSC. Date enrolled = date registered Royal Berkshire Hospital Only current attenders identified and registered Date enrolled = date registered Greenwich District Random recall and/or list of those identified through Hospital laboratory data at CDSC. Date enrolled = date registered Chester City Hospital Only current attenders identified and registered Date enrolled = date registered Huddersfield Royal Random recall and/or list of those identified through Infirmary laboratory data at CDSC. Date enrolled = date registered Central Middlesex All seroconverters identified through England & Wales Hospital Denominator Study. Not all have been reported, however. Date enrolled = date registered Lewisham Hospital Random recall and/or list of those identified through laboratory data at CDSC. Date enrolled = date registered North Middlesex Random recall and/or list of those identified through Hospital laboratory data at CDSC. Date enrolled = date registered Milton Keynes Hospital Random recall and/or list of those identified through laboratory data at CDSC. Date enrolled = date registered Barnsley District Random recall and/or list of those identified through General Hospital laboratory data at CDSC. Date enrolled = date registered Preston Royal Hospital Through comprehensive search by Manchester PHL of all seroconverters identified through Manchester and Preston PHLs and North Manchester Regional Virus Laboratory Date enrolled = seroconversion date Royal South Hampshire Random recall and/or list of those identified through Hospital laboratory data at CDSC. Date enrolled = date registered Upton Park Hospital Random recall and/or list of those identified through laboratory data at CDSC. Date enrolled = date registered ______
177 Clinical centre Extent to which all eligible seroconverters have been identified and included in the Register______King’s Mill Centre Random recall and/or list of those identified through laboratory data at CDSC. Date enrolled = date registered New Cross Hospital Random recall and/or list of those identified through laboratory data at CDSC. Date enrolled = date registered Drybum Hospital Random recall and/or list of those identified through laboratory data at CDSC. Date enrolled = date registered Ashford Hospital Random recall and/or list of those identified through laboratory data at CDSC. Date enrolled = date registered West Middlesex Hospital Random recall and/or list of those identified through laboratory data at CDSC. Date enrolled = date registered St Peter’s Hospital, Random recall and/or list of those identified through Chertsey laboratory data at CDSC. Date enrolled = date registered New east Surrey Random recall and/or list of those identified through Hospital laboratory data at CDSC. Date enrolled = date registered St Richard’s Hospital Random recall and/or list of those identified through laboratory data at CDSC. Date enrolled = date registered Basingstoke District Random recall and/or list of those identified through Hospital laboratory data at CDSC. Date enrolled = date registered High Wycombe General Random recall and/or list of those identified through Hospital laboratory data at CDSC. Date enrolled = date registered Worcester Royal Random recall and/or list of those identified through Infirmary laboratory data at CDSC. Date enrolled = date registered Walsall Manor Hospital Random recall and/or list of those identified through laboratory data at CDSC. Date enrolled = date registered Victoria Hospital, Through comprehensive search by Manchester PHL of all Blackpool seroconverters identified through Manchester and Preston PHLs and North Manchester Regional Virus Laboratory Date enrolled = seroconversion date
178 Clinical centre Extent to which all eligible seroconverters have been identified and included in the Register______Manchester Royal Through comprehensive search by Manchester PHL of all Infirmary seroconverters identified through Manchester and Preston PHLs and North Manchester Regional Virus Laboratory Date enrolled = seroconversion date Royal Oldham Hospital Through comprehensive search by Manchester PHL of all seroconverters identified through Manchester and Preston PHLs and North Manchester Regional Virus Laboratory Date enrolled = seroconversion date Royal Gwent Hospital Random recall and/or list of those identified through laboratory data at CDSC. Date enrolled = date registered Northampton General Random recall and/or list of those identified through Hospital laboratory data at CDSC. Date enrolled = date registered Chesterfield & N Random recall and/or list of those identified through Derbyshire Hospital laboratory data at CDSC. Date enrolled = date registered Middlesborough Random recall and/or list of those identified through Hospital laboratory data at CDSC. Date enrolled = date registered Essex County Hospital Random recall and/or list of those identified through laboratory data at CDSC. Date enrolled = date registered North Staffordshire Random recall and/or list of those identified through Hospital laboratory data at CDSC. Date enrolled = date registered Plymouth General Random recall and/or list of those identified through Hospital laboratory data at CDSC. Date enrolled = date registered Crosshouse Hospital Identified through Scottish Denominator study of all seroconverters in Scotland. Date enrolled = seroconversion date Glasgow Royal Identified through Scottish Denominator study of all Infirmary seroconverters in Scotland. Date enrolled = seroconversion date Aberdeen Royal Identified through Scottish Denominator study of all Infirmary seroconverters in Scotland. Date enrolled = seroconversion date Monklands Hospital Identified through Scottish Denominator study of all seroconverters in Scotland. Date enrolled = seroconversion date Glasgow Southern Identified through Scottish Denominator study of all General seroconverters in Scotland. Date enrolled = seroconversion date
179 Clinical centre Extent to which all eligible seroconverters have been identified and included in the Register______Ayrshire Central Identified through Scottish Denominator study of all Hospital seroconverters in Scotland. Date enrolled = seroconversion date Raigmore Hospital Identified through Scottish Denominator study of all seroconverters in Scotland. Date enrolled = seroconversion date Borders General Identified through Scottish Denominator study of all Hospital seroconverters in Scotland. Date enrolled = seroconversion date Luton & Dunstable Random recall and/or list of those identified through Hospital laboratory data at CDSC. Date enrolled = date registered Princess Margaret Random recall and/or list of those identified through Hospital, Swindon laboratory data at CDSC. Date enrolled = date registered Princess Royal Hospital, Random recall and/or list of those identified through Telford laboratory data at CDSC. Date enrolled = date registered Singleton Hospital Random recall and/or list of those identified through laboratory data at CDSC. Date enrolled = date registered Gloucester Royal Random recall and/or list of those identified through Hospital laboratory data at CDSC. Date enrolled = date registered Famham Road Hospital Random recall and/or list of those identified through laboratory data at CDSC. Date enrolled = date registered Baillie Street Clinic, Through comprehensive search by Manchester PHL of all Rochdale seroconverters identified through Manchester and Preston PHLs and North Manchester Regional Virus Laboratory Date enrolled = seroconversion date Whipps Cross Hospital Random recall and/or list of those identified through laboratory data at CDSC. Date enrolled = date registered Whittington Hospital Random recall and/or list of those identified through laboratory data at CDSC. Date enrolled = date registered Victoria Hospital, Identified through Scottish Denominator study of all Kirkcaldy seroconverters in Scotland. Date enrolled = seroconversion date Watford General Random recall and/or list of those identified through Hospital laboratory data at CDSC. Date enrolled - date registered
180 Appendix VI
Epidemiol. Infect. (1996), 117, 305-312 Copyright © 1996 Cambridge University Press
The UK Register of HTV Seroconverters: methods and analytical issues
\ UK Register of HIV Seroconverters (UKRHS) Steering Committee* I {Accepted 28 March 1996)
SUMMARY
A Register of HIV-infected persons who have had a negative antibody test within 3 years of their first antibody positive test (seroconverters) is being set up in the UK to monitor the distribution of times from HIV seroconversion to AIDS (the incubation period) and to death. It will also provide a national resource for use by those designing studies in this group of individuals. Clinicians caring for HIV-positive persons in Genito-Urinary Medicine, Infectious Disease and other departments throughout the UK were asked to participate by providing information on eligible subjects. Most laboratories undertaking HTV antibody testing were also contacted and asked to provide the name of the attending clinician for all seroconverters identified through the HTV laboratory reporting systems of the PHLS Communicable Disease Surveillance Centre (CDSC) and the Scottish Centre for Infection and Environmental Health (SCIEH) and for any other seroconverters known to them but not identified by CDSC or SCIEH. Data items sought for the Register include: sex, ethnic group, probable route of HTV transmission, annual CD4 counts, details of therapy and prophylaxis prescribed, AIDS-defining events and vital status. Follow up information is collected annually. Wherever possible, all seroconverters known to a clinic have been identified, whether currently alive or dead, either from clinic records or laboratory reporting or both. The objective is to establish and update a complete register of seroconverters on a long-term basis to provide reliable estimates of the incubation period on which future projections of AIDS cases in the UK can be made.
of future numbers of AIDS cases. These distributions BACKGROUND AND OBJECTIVES have changed since the beginning of the epidemic due Knowledge of the distribution of intervals from particularly to the introduction of anti-retroviral human immunodeficiency virus (HIV) infection to the treatment and prophylaxis for Pneumocystis carinii development of acquired immunodeficiency syndrome pneumonia and other opportunistic infections [1, 2]. (AIDS) and to death, and the factors affecting these It is likely that new advances in the management of intervals is vital for an understanding of the natural individuals with HIV infection will influence these history of HTV infection and for making projections distributions. Further, changes in the incubation • V. Berai, {Chair), ICRF Cancer Epidemiology Unit, Oxford; A. Babikerf, MRC HIV Clinical Trials Centre; UCLMS, London; R. P. Brettle, City Hospital, Edinburgh; C. Came, Addenbrooke’s HospitaL Cambridge; J. H. Darbyshiret, MRC HIV Clinical Trials Centre; UCLMS, London; B. G. Evanst, PHLS Communicable Disease Surveillance Centre, London; R. J. C. Gilson, Academic Dept of Sexually Transmitted Diseases, UCLMS, London; D. Goldberg, Scottish Centre for Infection and Environmental Health, Glasgow; D. A. Hawkins, Chelsea & Westminster Hospital, London ; D. Jeffries, St Bartholomew’s Hospital, London; A. M. Johnsonf, MRC UK HIV Epidemiology Co-ordinating Centre, UCLMS, London; M A. Johnson, Royal Free Hospital, School of Medicine, London; A. J. McMichael, Institute of Molecular Medicine, Oxford ; P. P. Mortimer, PHLS Central Public Health Laboratory, London ; A. N. Phillipsf, Royal Free Hospital School of Medicine, London; K. Porter{Project Co~ordinator), MRC HIV Clinical Trials Centre; UCLMS, London; A. Pozmak, King’s College Hospital, London; J. Weber, St Mary’s Hospital, London and S. Wellstecd UK Department of Health, London, t Members of the Executive Committee. Correspondence to; Kholoud Porter, MRC HIV Clinical Trials Centre. University College London Medical School, The Mortimer Market Centre, Mortimer Market, London WCIE 6AU, UK.
181 306 UKRHS Steering Committee period distribution could occur even in the absence of A maximum period of 3 years between the negative changes in available treatments and treatment uptake, and positive tests was chosen, but additional analyses due to the evolving distributions of new viral strains in using a narrower period will also be undertaken as HIV-infected persons (3,4], It is therefore important appropriate. Too narrow a period would inevitably to monitor whether there are changes in the in result in the selection o f a group o f individuals cubation period distribution and, if so, the extent of undergoing HIV testing at very frequent intervals. those changes and factors associated with them. A number of studies have provided estimates for Identifying study subjects these incubation periods in different population groups [5-22]. Most studies have tended to focus on Clinicians in the UK have reported AIDS cases to the one transmission category, e.g. homosexual men, Public Health Laboratory Service AIDS Centre at the injecting drug users, or haemophiliacs, are small in Communicable "Disease Surveillance Centre (CDSC) size, or are no longer recruiting new subjects. and the Scottish Centre for Infection and Environ A Register of HIV-infected individuals in whom the mental Health (SCIEH) since 1982 on a voluntary and date of seroconversion is known with reasonable confidential basis. Laboratory reporting to CDSC precision (seroconverters) is currently being set up in and SCIEH of newly-identified HIV infection began the UK to monitor changes in the distribution of in the latter part of 1984. AIDS case reporting is intervals from seroconversion to onset of AIDS, using estimated to be 87 % complete while the HTV the 1993 European case definition [23] (referred to as laboratory reporting is estimated to cover arotmd ' the incubation period’) and death. Seroconversion is 80 % of known infections [24,25]. Clinical and estimated as the mid-point between the last negative laboratory reporting nm independently but the use of and first HIV positive antibody test dates, and is used a soundex code (a code derived from the patient’s to approximate the time of infection with HIV. The surname comprising of a letter followed by three Register will also provide information on factors digits) as well as the date of birth allow both the link associated with the length of these intervals. It is between an HIV and an AIDS case report on the same intended to be complete and ongoing to provide individual to be made, and the elimination of possible constant and regular monitoring of these periods. The duplicates on either system. Register will be a shared national resource for use by those designing studies to help improve imderstanding Clinical centres of HIV pathogenesis and of immunological and Clinicians from Genito-Urinary Medicine, Infectious virological markers of infection and their relation to Disease and other specialities from centres in the UK prognosis. taking part in the Medical Research Council (MRC) clinical trials in HIV infection were invited to METHODS participate by providing information on patients who Definition of a seroconverter meet the definition of a seroconverter. Additionally, clinicians from centres not taking part in the MRC For the purposes of the Register, a seroconverter is trials but whose centre has reported a total of 10 or defined as an HIV-seropositive person aged 16 years more AIDS cases to CDSC or to SCIEH were also or over on whom an HIV antibody test was performed invited to participate. and found to be negative no more than 3 years prior to the first positive antibody test. In the absence of a Laboratories negative result, persons identified during the acute infection stage, namely when infection has occurred Laboratory directors and medical microbiologists but before full seroconversion, will also be included in were also contacted by letter and telephone and asked the study. Laboratory evidence for this acute infection to participate by helping to identify eligible subjects. is required, the details of which were reached by These include National Health Service, Public Health consensus among expert virologists at the request of and private laboratories, as well as regional Blood the Steering Committee (see Appendix). The Register Transfusion Centres. is not actively recruiting haemophiliac patients be In October 1986 the PHLS Collaborative Lab cause they are followed up by the UK Haemophilia oratory Study on HIV infection (the ' Denominator Centre Directors’ HIV Working Party. Study’) was initiated to gather information on all HIV
182 UK Register of HIV Seroconverters 307 testing and therefore identify repeat tests on the same the patient’s vital status and attendance at a clinic. As individual and hence incident infections. It comprised this is not always possible, information from the nine laboratories and expanded to include 18 labora clinic is also used; this may give biased estimates tories by the end o f 1989. As participating laboratories towards survivors if they include only current atten record all HIV tests, both negative and positive, the ders, but seroconverters with a negative test performed identification of seroconverters at these laboratories at another laboratory can also be identified. Com was made much simpler. A similar surveillance puterized records from the clinic are used whenever mechanism was set up in November 1988 in Scotland available, but in order to identify eligible subjects who through which all laboratories performing HIV testing had died or transferred to other centres before the report centrally to SCIEH. Colleagues at SCIEH have computerization of records had taken place the search also undertaken follow up, through the Blood is supplemented by case note review where feasible. Transfusion Service, of HIV seropositive donors with By close contact with centres and laboratories we previous HIV negative donations. aim to identify and minimize potential biases in the incubation period estimate (discussed below), which Detection of duplicate reports are due to the method of recruitment of eligible cases, by seeking information on the completeness of the The current CDSC HIV laboratory reporting form, local records, either electronic or manual, in particular which has been in use since March 1993, contains a the inclusion of patients no longer attending the specific question asking whether a negative result had centre or who had died. previously been obtained. For all reports where this Clinicians are also asked, with informed and item of information was provided, the reporting written consent from the patient, to take a blood laboratory was contacted and asked for the name and sample for central serum storage. Ethics committee address of the clinician looking after the patient, or, if approval was obtained for this from all participating this was not known, the name and address o f the clinical centres. clinician requesting the HIV test. The clinician was then asked to register the patient by completing a registration form and also to identify and register all Information collected other seroconverters from their centre including those The following information is collected for all sero known to be dead, transferred to other centres, or lost converters either by asking the clinician or research to follow-up. Duplicate reports made by the clinic and nurse to complete a form or by extraction from the the laboratory on the same individual were identified clinic database: Soundex code, initials, date of birth, by using soundex code and date o f birth matching. clinic/hospital number, sex, ethnic group, probable Prior to March 1993, there was no specific question route of virus transmission, likely country of infection, regarding a previous negative test on the CDSC form. presence of a seroconversion illness (with date and All laboratories who have reported new HIV infec main symptoms), dates of last negative and first HIV tions first diagnosed by them since January 1990 were positive antibody tests and where these were per also asked to check for any previous negative results formed, vital status and, if dead, date of death. for all individuals with a positive result between The following data are also collected initially and January 1990 and March 1993 who were not already on an annual basis: CD4 cell counts, details of known to have had a previous negative result. antiretroviral therapy and prophylaxis for oppor tunistic infections (either as open label therapy or as Retrospective ascertainment part o f a clinical trial), AIDS-defining events, vital A number of laboratories and clinical centres, par status and, if dead, date and likely cause(s). ticularly those with large case loads of HIV-infected Information about the nature and completeness of persons, include information on previous negative any existing local databases was sought from the results on their patient databases. In order to identify major centres caring for HIV infected individuals. On subjects eligible for the Register retrospectively, the basis of this, the study registration and follow up information is sought, wherever possible, from the forms were designed to be compatible with the local laboratory performing HIV testing. This method information collected routinely. This was of particular is least Likely to result in biased estimates of the relevance to information on drugs prescribed. At incubation period distribution as it is independent of registration, only information on the date of start of
183 308 UKRHS Steering Committee
Table 1. Characteristics of seroconverters reported to the Register: October 1994-February 1996
Number of reports Proportion of cases (n = 1348) (%) Sex Male 1203 89 Female 145 11 Exposure category Sex between men* 1040 77 Sex between men and 134 10 women Injecting drug use 146 11 Otber/undeterminedf 28 2 Seroconversion 1 calendar month 57 4 interval 1-12 months 688 51 13—24 months 382 28 25-36 months 221 16 Age at first antibody Median (range) 28(15-64) ' positive test (yrs)
* Includes 22 men who also reported to have injected drugs. t The Register is not actively recruiting haemophiliacs as they are already followed up by the UK Haemophilia Centre Directors’ HIV Working Party.
each drug prescribed is requested. Strict confi of deaths in the UK against those of all patients lost dentiality is maintained. No names or addresses of to follow up on the Register based on Soundex code, seroconverters are requested. The database for the date of birth and sex. As no names will be disclosed to study is held within a secure building with separate the Register by OPCS or GRO, once a possible match coded access and data are completely inaccessible or a number of probable matches are made they will except through the use of passwords known to key be verified with the collaboration of the clinician last individuals only. Data released to other researchers, known to have provided patient care. In this manner which would require the approval of the Steering the HIV status of persons on the Register will not Committee, would contain no information that could become disclosed to either OPCS or GRO. lead to the identification of seroconverters. RESULTS Follow up information By the end of February 1996, 142 clinical centres and Follow up data is collected on an annual basis 136 laboratories had been contacted, of which 131 through the appropriate clinical centre. Only in (92%) and 103 (76%), respectively, have agreed to formation on whether the patient is currently taking participate, representing 91 % of AIDS case reports the drug or any additional drugs prescribed or any and 97 % of HIV infection reports in the UK to date. stopped in the interval since last follow up will be A total of 2070 HIV antibody positive persons aged 16 sought. Any temporary changes in medication in the years or more who were not known to be haemophiliac intervening period will not be recorded as details are patients and had had a previous negative antibody unlikely to be available. It is unlikely that this level of test were identified from all sources. Full information detail would have a substantial impact on the analyses. has been received on 1526 individuals, 806 o f whom The AIDS databases both at CDSC and SCIEH will appear not to be known to either CDSC or SCIEH. also be used to identify AIDS cases and deaths The remaining 544 are known to the CDSC HIV occurring in patients who have stopped attending laboratory reporting system and to SCIEH and, as their clinic. clinical information has not yet been received, are not The Office of Population Censuses and Surveys included on the Register. (OPCS) in England and Wales and the General Of the 1526 seroconverters reported to the Register, Register Office (GRO) in Scotland are the central 987 (65%) were reported from 13 centres in London, registries for birth, marriage, and death. It is expected 219 (14%) from 12 centres in Scotland, and 320 that they will provide information annually on deaths (21 %) from 52 centres in the rest of the UK. Of these, not known to CDSC or SCIEH by matching records 1120 were identified through clinic records, 295
184 U K Register of HIV Seroconverters 309
Table 2. Year of estimated seroconversion* for casesvarious parametric models such as Weibull or G am m a reported October 1994-February 1996 distributions. The effects o f variables on the length o f survival will be examined using, principally, Cox Number of proportional hazard models. Year cases The aim is to include all seroconverters known to 1983 31 clinics and laboratories on the Register including 1984 49 patients alive at the time the Register is set up and all 1985 40 patients who have died. However, serious bias may 1986 55 1987 58 arise if the likelihood of a person being included on 1988 86 the Register is dependent on how rapidly that 1989 114 individual develops AIDS and/or dies. For example, 1990 155 in some centres seroconverters who have progressed 1991 168 rapidly and died may be less likely to be registered 1992 203 1993 214 than those who are currently attending. This problem 1994 134 arises when a clinic/laboratory does not have a 1995 40 complete record of all patients ever seen. In such 1996 1 centres data on, for example, a person who sero converted in 1987, and is registered in 1995, cannot be * Mid-point between the first positive and last negative HIV antibody tests. used to derive information on the first 8 years of the NB. not all individuals will contribute information to incubation period since, if the patient had died within estimates of the incubation period from this point (see 8 years (i.e. before 1995), they may well have not been Discussion - Statistical issues and possible biases). included on the Register. Such a patient can thus only be entered in the ‘risk set’ for the analysis in 1995 at through laboratory records, and 111 through both the 8 years. This is referred to as ‘late entry’. If such a clinic and laboratory. The seroconversion interval, i.e. patient had attended at a clinic or was identified the time difference between a first antibody positive through a laboratory with complete records of every and last antibody negative tests for HIV, is within one patient seen/tested since, for example 1990, they can calendar month for 57 reports, 1-12 months for 688 be entered into the risk set at 3 years from sero reports, 13-24 months for 382 reports, 25-36 months conversion (1990-1987 = 3 years) since survival or for 221 reports and over 36 months for 153 reports. death after this date will not have influenced whether For 25 reports no date was given for either the they are included on the Register. positive or negative tests or both. For individuals from centres whose ascertainment The characteristics of the 1348 reports with a is believed to be complete from laboratory and/or seroconversion interval of 3 years or less are outlined clinic records (that is, all patients alive, dead, in Table 1. Most (77%) were infected through sex transferred and lost to follow up are included), follow between men, 10% through sex between men and up through the Register commences from each women, and 11 % through injecting drug use. This individual’s seroconversion time. However, for centres contrasts with 65%, 19% and 12%, respectively, where ascertainment is judged to be complete only among reported cases of HIV infection in the UK [26]. from a given date, h, seroconverters are treated as Most séroconversions are estimated to have taken ‘late entry’ subjects and each enters the ‘risk set’ at {h place in 1992 and 1993 but information is available on - estimated date of seroconversion). If ascertainment persons believed to have seroconverted since 1983 from a centre is not complete from any point in time, (Table 2). follow up for all seroconverters from that centre will commence from the date of entry to the Register. DISCUSSION Follow up (i.e. entry into the ‘risk set’) for subjects already known to be HIV infected when transferring Statistical issues and possible biases into a centre commences from the date they first The incubation period distribution, given by the attended the clinic with complete records on all cumulative probability of developing AIDS within t patients. If this is uncertain, or if neither clinic have years o f infection will be estimated using Kaplan- complete records the follow up of such subjects Meier estimation. Attempts will also be made to fit commences from date of entry to the Register.
185 310 UKRHS Steering Committee
Table 3. Number of patients recruited from each of 3 hypothetical clinics
Estimated year of seroconversion Total number Clinic 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 recruited
A 20* 10 0 20 0 25 0 0 20 0 5 100 B 0 25 0 0 0 25 0* 0 25 0 25 100 C 25 25 0 25 0* 0 0 0 25 0 0 100
* Time h from which records are believed to be complete. Patient numbers to the left of these are treated as ‘late entry’ subjects in the analysis.
* T able 4. Life table of retrospectively identified seroconverters
Cumulative Time from event-free seroconversion survival (yrs) Numbers of patients in risk set Number of events probability
0 175* 10 0-94 1 165 + 25t + 25t 10 0-90 2 205 15 0-83 3 190+251 6 0-81 4 209 + 251 10 0-78 5 224 + 251 12 0-74 6 237 7 0-72 7 230 10 0-69 8 220 14 0-64 9 206 13 0-60
* 100 patients from clinic A, 50 from clinic B, and 25 from clinic C. t Late entry subjects from clinic B. t Late entry subjects from clinic C.
To give examples o f this potential bias, consider the For patients in chnic B follow up can only start following scenarios (Table 3): from January 1990 since information is incomplete for (i) In cUnic A, it is possible to identify all HIV the period prior to this date and their follow up infected persons who have ever attended the cUnic commences from that point so that they enter the risk regardless of whether they have since died, transferred set at the number of years after seroconversion they to other centres or are currently alive and attending at were in 1990. Similarly, patients registered from clinic this clinic. C enter the risk set at the number o f years after (ii) In clinic B, records became computerized for seroconversion they were in 1988 and their follow up patients attending from January 1990 onwards. begins from that point. To illustrate how the problem Therefore, eligible subjects who had died or trans o f ‘late entry’ is dealt with, let us suppose that 100 ferred to other centres prior to January 1990 were not eUgible seroconverters are identified from each of the entered onto the system and cannot, therefore, be three centres A, B and C and their estimated dates of identified. HIV seroconversion are as illustrated in Table 3. The (iii) In clinic C, complete records for attenders are resulting life table for analysis would resemble Table 4. available from January 1988, and so patients who had A bias concerning the representativeness of sero died or transferred to other centres prior to that date converters on the Register may arise because persons are not included. who undergo repeat testing for HIV are not likely to For patients in clinic A follow up can be backdated be truly representative of all HIV infected persons. to the date of their individual HIV positive tests and For example, heterosexuals acquiring the virus their date of entering the risk set is, therefore, the through sexual contact often have not previously estimated seroconversion date. perceived themselves to be at risk of HIV infection
186 UK Register of HIV Seroconverters 311 and so tend not to have had previous negative tests for recently acquired infection, as detailed below. When HIV and, therefore, are under-represented among the results o f the investigations are available, indi seroconverters. viduals not meeting these criteria will be removed from the Register. (b) Any other HIV positive individual without a Future developments previous negative test in whom the laboratory criteria As persons with a previous negative antibody test are below have been met will be enrolled. likely to remain a small proportion of all HIV-infected individuals, international collaboration is important Laboratory criteria of recent seroconversion (one in order to pool information and increase the degree criterion required) of confidence in estimates of the incubation period, not only for homosexual and bisexual men but also (a) The presence o f p24 antigen, confirmed by for heterosexual men and women. This is of crucial neutralization, and/or PCR positivity, in the absence importance for making reliable estimates o f current of total anti-HIV at that time, but in whom anti-HIV and future AIDS case load and recent HIV prevalence is demonstrated in a specimen collected within 6 [24, 25]. months. Due to the relatively low incidence of HIV infection, (b) Weak reactivity in EIA or low titre ( < 256) by incident cohorts, i.e. cohorts of initially HIV sero Serodia-HIV and IgM anti-HIV or p24 antigen and negative persons followed up through seroconversion seroconversion-type Western Blot pattern, with to AIDS and death, require many years of follow up. either; A number of studies on the incubation period - increasing titres o f anti-HIV /E IA reactivity in speci overcome this problem by using information from mens collected within a 3-month period but tested prevalent cohorts [6,8,11,12,27,28], i.e. persons simultaneously, or already known to be HIV seropositive when recruited. -development of new bands or intensification of As subjects in prevalent cohorts are by definition weakly reactive bands in two or more sequential already HIV infected, the unobserved period o f their sera collected at an interval of no more than 3 HIV infection has to be accounted for and supple months and tested contemporaneously by Western mented with information from other sources. The UK Blot. Register of HIV Seroconverters aims to include all seroconverters, to identify all potential biases and to ACKNOWLEDGEMENTS account for them in the analysis. This will ensure that the Register is as close as possible to a cohort of We would like to acknowledge the work of our individuals with incident HIV infection. colleagues in the clinics and laboratories, too many to Response from laboratories and clinical centres in list, who are collaborating in the study by identifying the UK has been encouraging. Much of the effort in eligible subjects and providing information on them. identifying eligible subjects is concerned with retro A list of those participating and further information spective identification and ascertainment of the com on the Register may be obtained from the Project Co pleteness of the records in individual clinical centres. ordinator. The mechanism for identifying seroconverters pro- We would also like to thank colleagues at CDSC spectively is unlikely to require as much effort. (Neil Macdonald, Pauline Kaye), SCIEH (Jim McMenamin, Dave Reynolds, Glenn Codere, Ahilya Noone), and the MRC HIV Clinical Trials Centre APPENDIX (Charlotte Duff) for their help with this study. The Definition of ‘acute infection’ Register is funded by a grant from the Medical Research Council of the United Kingdom. (a) An individual believed to be at increased risk for HIV infection presenting with an illness compatible with acute primary HIV infection without docu REFERENCES mentation of a negative HIV antibody test will be 1. Hessol NA, Koblin BA, van Griensven GJP, et al. enrolled into the study pending the outcome of Progression of human immunodeficiency virus type I laboratory investigations to establish evidence of (HIV-1) infection among homosexual men in hepatitis
187 312 UKRHS Steering Committee
B vaccine triai cohorts in Amsterdam, New York City, infected by HTV through blood transfusion. AIDS and San Francisco, 1978-1991. Am J Epidemiol 1994; 1990; 4: 125-9. 139: 1077-87. 15. Pedersen C, Lindhart B0, Jensen BL, et aL Clinical 2. Munoz A, Schrager LK, Bacellar H, et al. Trends in the course of primary infection: consequences for sub incidence of outcomes defining acquired immuno sequent course of infection. BMJ 1989; 299: 154-7. deficiency syndrome (AIDS) in the Multicentre AIDS 16. Darby SC, Rizza CR, Doll R, et al. Incidence of AIDS Cohort Study: 1985-1991. Am J Epidemiol 1993: 137: and excess of mortality associated with HIV in 423-38. haemophiliacs in the United Kingdom: report on behalf 3. Learmont J, Tindall B, Evans L, et al. Long-term of the directors of haemophilia centres in the United symptomless HIV-1 infection in recipients of blood Kingdom. BMJ 1989; 298: 1064-8. products from a single donor. Lancet 1992; 340: 863-7. 17. Phillips AN, Sabin CA, Elford J, et al. Use of CD4 4. Ward JW, Bush TJ, Perkins HA, et al. The natural lymphocyte count to predict long term survival free of AIDS after HIV infection. BMJ 1994; 309: 309-13. history of transfusion-associated infection with human 18. Brettle RP, McNeil AJ, Gore SM, et aL The Edinburgh immunodeficiency virus. New Engl J Med 1989; 321: City Hospital cohort analysis of enrolment, progression 947-52. and mortality by baseline covariates. QJM 1995; 88 (7): 5. Rosenberg PS, Goedert JJ, Biggar RJ, et al. Effect of 479-91. age at seroconversion on the natural AIDS incubation 19. Msellati P, Dupon M, Morlat P, et aL A cohort of 89 distribution. AIDS 1994; 8: 803-10. HIV-1 infected adult patients contaminated by blood 6. Saah AJ, Munoz A, Kuo V, et al. Predictors of the risk products: Bordeaux 1981-1989. AIDS 1990; 4:1105-9. of development of Acquired Immunodeficiency Syn 20. Chiarotti F, Palombi M, Schinaia N, et aL Median time drome within 24 months among gay men seropositive from seroconversion to AIDS in Italian HIV-positive for human immunodeficiency virus type I: a report haemophiliacs : different parametric estimates. Stat Med from the Multicenter AIDS Cohort Study. Am J 1994; 13: 163-75. Epidemiol 1992; 135: 1147-55. 21. Lepri AC, Pezzotti P, Domicci M, et al. HTV disease 7. Margolick JB, Munoz A, Vlahov D, et al. Changes in progression in 854 women and men infected through T-lymphocyte subsets in intravenous drug users with injecting drug use and heterosexual sex and followed for HIV-1 infection. JAMA 1992; 267: 1631-6. up to nine years from seroconversion. Br Med J 1994; 8. Lang W, Perkins H, Anderson RE, et al. Patterns of T 309: 1537-^2. lymphocyte changes with human immunodeficiency 22. Veugelers PJ, Page KA, Tindall B, et al. Determinants virus infection: from seroconversion to the development of HIV disease progression among homosexual men of AIDS. J Acq Immun Defic Synd 1989; 2: 63-9. registered in the tricontinental seroconverter study. Am 9. Hessol NA, Lifson AR, O’Malley PM, et al. Prevalence, J Epidemiol 1994; 140 : 747-58. incidence and progression of human immunodeficiency 23. Ancelle-Park RA. European AIDS definition. Lancet virus infection in homosexual and bisexual men in 1992 ; 339: 671. hepatitis B vaccine trials, 1978-1988. Am J Epidemiol 24. Report of an expert group (Chairman, Professor NE 1989; 130: 1167-75. Day) convened by the Director of the Public Health 10. Gorham ED, Garland PC, Mayers DL, et al. CD4 Laboratory Service on behalf of the Chief Medical Officers. The incidence and prevalence of AIDS and lymphocyte counts within 24 months of human im prevalence of other severe HIV disease in England and munodeficiency virus seroconversion. Arch Intern Med Wales for 1995-1999: projections using data to the end 1993; 153: 869-76. of 1994. CDR 1996; 6. 11. Schechter MT, Boyko WJ, Craib KJP, et al. Effects of 25. Report of a Working Group convened by the Chief long-term seropositivity to human immunodeficiency Medical Officer for Scotland December 1995. AIDS virus in a cohort of homosexual men. AIDS 1987; 1: and severe HIV-related disease in Scotland. Predictions 77-82. to the end of 1999. ANSWER, 29 December 1995. 12. Burcham J, Marmor M, Dubin N, et al. CD4% is the 26. AIDS and HIV-1 infection in the United Kingdom: best predictor of development of AIDS in a cohort of monthly report. CDR, 1996; 6: 25-8. HIV-infected homosexual men. AIDS 1991 ; 5: 365-72. 27. Hendriks JCM, Medley GF, van Griensven GJP, et al. 13. Keet IPM, Krijnen P, Koot M, et al. Predictors of rapid The treatment-free incubation period of AIDS in a progression to AIDS in HIV-1 seroconverters. AIDS cohort of homosexual men. AIDS 1993; 7: 231-9. 1993; 7: 51-7. 28. Kuo J-M, Taylor JM G, Detels R. Estimating the AIDS 14. Blaxhult A, Granath F, Lidman K, Giesecke J. The incubation period from a prevalent cohort. Am J influence of age on the latency period to AIDS in people Epidemiol 1991 ; 133: 1050-7.
188 REFERENCES
1. Centers for Disease Control. Pneumocystis pneumonia- Los Angeles. MMWR 1981;30:250-2. 2. Centers for Disease Control. Kaposi’s sarcoma andPneumocystis pneumonia among homosexual men- New York City and California. MMWR 1981;30:305-8. 3. Walzer PD, Perl DP, Krogstad DG, Rawson PG, Schultz MG. Pneumocystis carinii pneumonia in the United States. Epidemiologic, diagnostic and clinical features. Ann Intern Med 1974;80:83-93. 4. Centers for Disease Control. CDC task force on Kaposi’s sarcoma and opportunistic infections.N Engl J Med 1982;306:248-52. 5. Gange RW, Jones EW. Kaposi’s sarcoma and immunosuppressive therapy: an appraisal. Clin Exp Dermatol 1978;3:135-46. 6. Safai B, Good RA. Kaposi’s sarcoma: a review and recent developments. CA Cancer J Clin 1981;31:1-12. 7. Goedert JJ, Neuland CY, Wallen WC, et al. Amyl nitrite may alter T lymphocytes in homosexual men.Lancet 1982;1:412-6. 8. Sonnabend J, Witkin SS, Purtilo DT. Acquired immunodeficiency syndrome, opportunistic infections, and malignancies in male homosexuals. A hypothesis of etiologic factors in pathogenesis.JAMA 1983;249:2370-4. 9. Durack DT. Opportunistic infections and Kaposi’s sarcoma in homosexual men.N EnglJM ed 1981;305:1465-7. 10. Centers for Disease Control. Update on acquired immune deficiency syndrome (AIDS) - United States. MMWR 1982;31:507-14. 11. Centers for Disease Control. Pneumocytsis carinii pneumonia among persons with hemophilia A. MMWR 1982;31:365-7. 12. Poon MC, Landay A, Prasthofer EF, Stagno S. Acquired immunodeficiency syndrome with Pneumocystis carinii pneumonia and Mycobacterium avium- intracellulare infection in a previously healthy patient with classic hemophilia. Clinical, immunologic, and virologie findings.Ann Intern Med 1983;98:287-90. 13. Elliott JL, Hoppes WL, Platt MS,et al. The acquired immunodeficiency syndrome and Mycobacterium avium-intracellulare bacteremia in a patient with hemophilia. Ann Intern Med 1983;98:290-3.
189 14. Davis KC, Horsburgh CR Jr, Hasiba U, Schocket AL, Kirkpatrick CH. Acquired immunodeficiency syndrome in a patient with hemophilia. Ann Intern Med 1983;98:284-6. 15. Centers for Disease Control. Possible transfusion-associated acquired immune deficiency syndrome (AIDS) - California. MMWR 1982;31:652-4. 16. Centres for Disease Control. Immunodeficiency among female sex partners of males with acquired immunodeficiency syndrome (AIDS) - New York. MMWR 1983;31:697-8. 17. deShazo RD, Andes WA, Nordberg J, et al. An immunologic evaluation of hemophiliac patients and their wives. Relationships to the acquired immunodeficiency syndrome. Ann Intern Med 1983;99:159-64. 18. Masur H, Michelis MA, Wormser GF, et al. Opportunistic infection in previously healthy women. Initial manifestations of a community-acquired cellular immunodeficiency. Ann Intern Med 1982;7:533-9. 19. Harris C, Small CB, Klein RS, et al. Immunodeficiency in female sexual partners of men with the acquired immunodeficiency syndrome. N Engl J Med 1983;308:1181-4. 20. Rubinstein A, Sicklick M, Gupta A, et al. Acquired immunodeficiency with reversed T4/T8 ratios in infants bom to promiscuous and dmg-addicted mothers. JAMA 1983;249:2350-6. 21. Oleske J, Minnefor A, Cooper R Jr, et al. Immune deficiency syndrome in children. 1983;249:2345-9. 22. Ammann AJ, Cowan MJ, Wara DW, et al. Acquired immunodeficiency in an infant: possible transmission by means of blood products. Lancet 1983;1:956-8. 23. Du Bois RM, Branthwaite MA, Mikhail JR, Batten JC. Primary Pneumocystis carinii and cytomegalovims infection. Lancet 1981;ii:1339. 24. CDSC. The surveillance of Kaposi’s sarcoma and opportunistic infections in homosexual males in England and Wales. CDR 1982;34:3-4. 25. Auerbach DM, Darrow WW, Jaffe HW, Curran JW. Cluster of cases of the acquired immune deficiency syndrome: patients linked by sexual contact. Am J Med 1984;76:487-92. 26. Barré-Sinoussi F, Chermann JC, Rey F, Nugeyre MT, Chamaret S, Gmest J, et al. Isolation of a T-lymphotropic retrovims from a patient at risk for acquired immune deficiency syndrome (AIDS). Science 1983;220:868-71.
190 27. Gallo RC, Sarin PS, Gelmann EP, et al. Isolation of human T-cell leukaemia virus in acquired immune deficiency syndrome (AIDS). Science 1983;224:500-3. 28. Coffin J, Haase A, Levy JA, Montagnier L, et al. Human immunodeficiency viruses (letter). Science 1986;232:697. 29. McDougal JS, Kennedy MS, Sligh JM, et al. Binding of the HTLV-III/LAV to T4+ T cells by a complex of the 110k molecule and the T4 molecule. Science 1985;231:382-5. 30. Centers for Disease Control. Current trends: classification system for human T lymphotropic virus type Ill/lymphadenopathy associated virus infections. MMWR 1986; 35:334-9. 31. Centers for Disease Control. 1993 revised classification system for HIV infection and expanded surveillance case definition for AIDS among adolescents and adults. MMWR 1992;41(RR17):1-19. 32. Ho DD, Samgadharan MG, Resnick L, et al. Primary human T-lymphotropie virus type III infection. Ann Intern Med 1985;103:880-3. 33. Tindall B, Cooper DA. Primary HIV infection: host responses and intervention strategies. AIDS 1991;5:1-14. 34. Centers for Disease Control. Update on Kaposi’s sarcoma and opportunistic infections in previously healthy persons- United States. MMWR 1982;31:294- 301. 35. Centers for Disease Control. Revision of the case definition of acquired immunodeficiency syndrome for national reporting- United States. MMWR 1985;34:373-5. 36. Centers for Disease Control. Revision of the CDC surveillance case definition for acquired immunodeficiency syndrome. MMWR 1987;36:1S-15S. 37. Anonymous. New case definition. 1993;341:441. 38. Report of an expert group (Chairman Professor NE Day) convened by the Director of the Public Health Laboratory Service on behalf of the Chief Medical Officers. The incidence and prevalence of AIDS and prevalence of other severe HIV disease in England and Wales for 1995-1999: projections using data to the end of 1994. CDR 1996;6(review no. 1). 39. Report of a Working Group convened by the Chief Medical Officer for Scotland December 1995. AIDS and severe HIV-related disease in Scotland. Predictions to the end of 1999. ANSWER 29 December 1995.
191 40. Lui K-J, Lawrence DN, Morgan WM, Peterman TA, Haverkos HW, Bregman DJ. A model-based approach for estimating the mean incubation period of transfusion- associated acquired immunodeficiency syndrome. Proc Natl Acad Soi 1986; 83: 3051-5. 41. Bacchetti P, Jewell NP. Nonparametric estimation of the incubation period of AIDS based on a prevalent cohort with unknown infection times. Biometrics 1991;47:947-60. 42. Kuo JM, Taylor JM, Detels R. Estimating the AIDS incubation period from a prevalent cohort.Am J Epidemiol 1991;133:1050-7. 43. Coates RA, Farewell VT, Raboud J, et al. Cofactors of progression to acquired immunodeficiency syndrome in a cohort of male sexual contacts of men with human immunodeficiency virus disease. Am J Epidemiol 1990;,132:717-22. 44. Page Shafer K, Delorenze GN, Satariano WA, Winkelstein W Jr. Comorbidity and survival in HIV infected men in the San Francisco Men's Health Survey. Ann Epidemiol 1996;6:420-30. 45. Pedersen C, Kolby P, Sindrup J, et al. The development of AIDS or AIDS related conditions in a cohort of HIV antibody positive homosexual men during a 3 year follow up period.J Intern Med 1989;225:403-9. 46. Stites DP, Moss AR, Bacchetti P, et al. Lymphocyte subset analysis to predict progression to AIDS in a cohort of homosexual men in San Francisco.Clin Immunol Immunopathol 1989;52:96-103. 47. Moss AR, Bacchetti P, Osmond D, et al. Seropositivity for HIV and the development of AIDS or AIDS related condition: three year follow up of the San Francisco General Hospital cohort.Br Med J Clin Res Ed 1988;296:745-50. 48. Tindall B, Cooper DA, Donovan B, Barnes T, Philpot CR, Gold J, Penny R. The Sydney AIDS Project: development of acquired immunodeficiency syndrome in a group of HIV seropositive homosexual men.AustNZJ Med 1988;18:8-15. 49. Polk BF, Fox R, Brookmeyer R, et al. Predictors of the acquired immunodeficiency syndrome developing in a cohort of seropositive homosexual men. N EnglJ Med 1987;316:61-6. 50. Chmiel JS, Detels R, Kaslow RA, Van Raden M, Kingsley LA, Brookmeyer R Factors associated with prevalent human immunodeficiency virus (HIV) infection in the Multicenter AIDS Cohort Study. Am J Epidemiol 1987;126:568-77.
192 51. Brookmeyer R, Gail MH, Polk BF. The prevalent cohort study and the acquired immunodeficiency syndrome. Am J Epidemiol 1987;126:14-24. 52. Weber JN, Wadsworth J, Rogers LA, et al. Three year prospective study of HTLV III/LAV infection in homosexual men.Lancet 1986;i:l 179-82. 53. Tindall B, Cooper DA, Burcham J, Gold J, Penny R. Clinical and immunologic sequelae of AIDS retrovirus infection.Aust NZJMed 1986;,16:749-56. 54. Vittinghoff E, Malani HM, Jewell NP. Estimating patterns of CD4 lymphocyte decline using data from a prevalent cohort of HIV infected individuals. Stat Med 1994;13:1101-18 55. Alcabes P, Selwyn PA, Davenny K, et al. Laboratory markers and the risk of developing HIV 1 disease among injecting drug users .4/D5' 1994;8:107-15. 56. Alcabes P, Mufioz A, Vlahov D, Friedland G. Maturity of human immunodeficiency virus infection and incubation period of acquired immunodeficiency syndrome in injecting drug users. Ann Epidemiol 1994;4:17- 26. 57. Margolick JB, Mufioz A, Vlahov D, et al. Direct comparison of the relationship between clinical outcome and change in CD4+ lymphocytes in human immunodeficiency virus positive homosexual men and injecting drug users. Arch Intern Med 1994;154:869-75. 58. Mufioz A, Carey V, Taylor JM, et al. Estimation of time since exposure for a prevalent cohort.Stat Med 1992;11:939-52. 59. Margolick JB, Mufioz A, Vlahov D, Solomon L, Astemborski J, Cohn S, Nelson KE. Changes in T lymphocyte subsets in intravenous drug users with HIV 1 infection. JAMA 1992;267:1631-6. 60. Lau RK, Hill A, Jenkins P, et al. Eight year prospective study of HIV infection in a cohort of homosexual men clinical progression, immunological and virological markers. Int J STD AIDS 1992;3:261-6. 61. Taylor JMG, Schwartz K, Detels R. The time from infection with Human Immunodeficiency Virus (HIV) to the onset of AIDS.JInfec Dis 1986;154:694-7. 62. Kuo J-M, Taylor JMG, Detels R. Estimating the AIDS incubation period from a prevalent cohort.Am J Epidemiol 1991;133:1050-7. 63. Hessol NA, Lifson AR, O’Malley PM, Doll LS, Jaffe HW, Rutherford GW. Prevalence, incidence, and progression of human immunodeficiency virus
193 infection in homosexual and bisexual men in hepatitis B vaccine trials, 1978- 1988. Am J Epidemiol 1989;130:1167-75. 64. Rutherford GW, Lifson AR, Hessol NA, et al. Course of HIV-1 infection in a cohort of homosexual and bisexual men: an 11 year follow up study. BMJ 1990;301:1183-8.
65. Lifson AR, Hessol NA, Buchbinder SP, et al. Serum P2 -microglobulin and prediction of progression to AIDS in HIV infection.Lancet 1992;339:1435-40. 66. Buchbinder SP, Katz MH, Hessol NA, O’Malley PM, Holmberg SD. Long-term HIV-1 infection without immunologic progression.AIDS 1994;8:1123-8. 67. Hendriks JCM, Medley GF, van Griensven GJP, Coutinho R, Heisterkamp SH, van Druten HAM. The treatment free incubation period of AIDS in a cohort of homosexual men. 1993;7:231-9. 68. Keet IPM, Krijnen P, Koot M, Lange JMA, Miedema F, Goudsmit J, Coutinho RA. Predictors of rapid progression to AIDS in HIV-1 seroconverters.AIDS 1993;7:51-7. 69. Hessol NA, Koblin BA, van Griensven GJP, et al. Progression of human immunodeficiency virus type 1 (HIV-1) infection among homosexual men in hepatitis B vaccine trial cohorts in Amsterdam, New York City, and San Francisco, 1976-1991. Am J Epidemiol 1994;139:1077-87. 70. Schechter MT, Craib KJP, Le TN, et al. Progression to AIDS and predictors of AIDS in seroprevalent and seroincident cohorts of homosexual men.AIDS 1989;3:347-53. 71. Schechter MT, Craib KJP, Le TN, et al. Susceptibility to AIDS progression appears early in HIV infection. AIDS 1990;4:185-90. 72. Hogg RS, Strathdee SA, Craib KJP, O’Shaughnessy MV, Montaner JSG, Schechter MT. Lower socioeconomic status and shorter survival following HIV infection. Lancet 1994;344:1120-4. 73. Munoz A, Wang M-C, Bass S, et al. Acquired immune deficiency syndrome (AIDS)-firee time after human immunodeficiency virus type 1 seroconversion in homosexual men. Am J Epidemiol 1989;130:530-9. 74. Phair J, Jacobson L, Detels R, et al. Acquired immune deficiency syndrome occurring within 5 years of infection with human immunodeficiency virus type-1: The Multicenter AIDS Cohort Study. J Acquir Imm Defic Syndr 1992;5:490-6.
194 75. Taylor JMG, Kuo J-K, Detels R. Is the incubation period of AIDS lengthening? J Acquir Imm DefSynd 1991;4:69-75. 76. Veugelers PJ, Page KA, Tindall B, et al. Determinants of HIV disease progression among homosexual men registered in the Tricontinental Seroconverter Study.Am J Epidemiol 1994;140:747-58. 77. Giesecke J, Scalia-Tomba G, Hakansson C, Karlsson A, Lidman K. Incubation time of AIDS: progression of disease in a cohort of HIV-infected homosexual and bisexual men with known dates of infection. Scand J Infect Dis 1990;22:407-11. 78. Darby SC, Rizza CR, Doll R, Spooner RJD, Stratton IM, Thakrar B. Incidence of AIDS and excess mortality associated with HIV in haemophiliacs in the United Kingdom: report on behalf of the directors of haemophilia centres in the United Kingdom. R M /1989;298:1064-8. 79. Darby SC, Doll R, Thakrar B, Rizza CR, Cox DR. Time from infection with HIV to onset of AIDS in patients with haemophilia in the UK. Stat Med 1990;298:1064-8. 80. Phillips AN, Lee CA, Elford J, et al. More rapid progression to AIDS in older HIV infected people: the role of CD4+ T-cell counts. J Acquir Immune Defic Syndr 1991;4:970-5. 81. Lee CA, Phillips AN, Elford J, et al. Progression of HIV disease in a haemophiliac cohort followed for eleven years and the effect of treatment.BMJ 1991;303:1093-6. 82. Goedert JJ, Kessler CM, Aledort LM, et al. A prospective study of human immunodeficiency virus type 1 infection and the development of AIDS in subjects with haemophilia. 1989;321:1141-8. 83. Eyster ME, Gail MH, Ballard JO, Al-Mondhiry H, Goedert JJ. Natural history of human immunodeficiency virus infections in haemophiliacs: Effects of T-cell subsets, platelet counts, and age. Ann Int Med 1987;107:1-6. 84. Ragni MV, Kingsley LA. Cumulative risk for AIDS and other HIV outcomes in a cohort of haemophiliacs in Western Pennsylvania. J Acquir Imm Defic 1990;3:708-13. 85. Schinaia N, Ghirardini A, Chiarotti F, Gringeri A, Mannucci PM. Progression to AIDS among Italian HIV-seropositive haemophiliacs.AIDS 1991;5:385-91.
195 86. Chiarotti F, Palombi M, Schinaia N, Ghirardini A, Bellocco R. Median time from seroconversion to AIDS in Italian HIV-positive haemophiliacs: Different parametric estimates. Stat Med 1994;13:163-75. 87. Medley GF, Anderson RM, Cox DR, Billard L. Incubation period of AIDS in patients infected via blood transfusion. Nature 1987;328:719-21. 88. Ward JW, Bush TJ, Perkins HA, et al. The natural history of transfusion-associated infection with human immunodeficiency vims. New EnglJ Med 1989; 321:947-52. 89. Msellati P, Dupon M, Morlat P, Lacoste D, Pellegrin J, Dabis F. A cohort study of 89 HIV-1 infected adult patients contaminated by blood products: Bordeaux 1981- 1989. 1990;4:1105-9. 90. Blaxhult A, Fredrik G, Lidman K, Giesecke J. The influence of age on the latency period to AIDS in people infected by HIV through blood transfusion. AIDS 1990;4:125-9. 91. Downs AM, Ancelle-Park RA, Costagliola D,et al. Trans fusion-associated AIDS cases in Europe: estimation of the incubation period distribution and prediction of future cases. Acquir Imm DefSynd 1991;4:805-13. 92. Kopec-Schrader E, Tindall B, Learmont J, Wylie B, Kaldor JM. Development of AIDS in people with transfusion-acquired HIV infection.AIDS 1993;7:1009-13. 93. Rezza G, Lazzarin A, Angarano G, et al. The natural history of HIV infection in injecting dmg users: risk of disease progression in a cohort of seroconverters. ÆD5" 1989;3:87-90. 94. The Italian Seroconversion Study. Disease progression and early predictors of AIDS in HIV-seroconverted injecting dmg users. AIDS 1992;6:421-6. 95. Flegg PJ. The natural history of HIV infection: A study in Edinburgh dmg users. J Infection 1994;29:311-21. 96. Jason J, Lui K-J, Ragni MV, Hessol NA, Darrow WW. Risk of developing AIDS in HIV-infected cohorts of haemophiliac and homosexual men. JAMA 1989;261:725-7. 97. Biggar RJ. AIDS incubation in 1891 HIV seroconverters from different exposure groups. AIDS 1990;4:1059-66. 98. Rosenberg PS, Goedert JJ, Biggar RJ for the Multicenter Hemophilia Cohort Study and the International Registry of Seroconverters. Effect of age at seroconversion on the natural AIDS incubation distribution.AIDS 1994;8:803-10.
196 99. Carré N, Deveau C, Belanger F, et al. Effect of age and exposure group on the onset of AIDS in heterosexual and homosexual HIV-infected patients. AIDS 1994;8:797-802. 100. Rezza G, Lazzarin A, Angarano G, et al. Risk of AIDS in HIV seroconverters: a comparison between intravenous drug users and homosexual males. Eur J Epidemiol 1990;6:99-101. 101. Sinicco A, Fora R, Sciandra M, Luchini A, Caramello P, Gioannini P. Risk of developing AIDS after primary acute HIV-1 infection. J Acquir Imm Defic Syndr 1993;6:575-81. 102. Giesecke J, Scalia-Tomba G, Berglund 0, Bemtorp E, Schulman S, Stigendal L. Incidence of symptoms and AIDS in 146 Swedish haemophiliacs and blood transfusion recipients infected with human immunodeficiency virus. BMJ 1988;297:99-103. 103. Operskalski EA, Stram DO, Lee H, et al. Human immunodeficiency virus type 1 infection: relationship of risk group and age to rate of progression to AIDS. Transfusion Safety Study Group. J Infect Dis 1995;172:648-55. 104. Cozzi Lepri A, Pezzotti P, Dorrucci M, Phillips AN, Rezza G. HIV disease progression in 854 women and men infected through injecting drug use and heterosexual sex and followed for up to nine years from seroconversion.BMJ 1994;309:1537-42. 105. Waight PA, Rush AM, Miller E. Surveillance of HIV infection by voluntary testing in England. Commun Dis Rep CDR Rev 1992,2:R85-90. 106. Goldberg DJ, Emslie JA, Smyth W, Reid D. A system for surveillance of voluntary HIV testing: results of the first 2 years, 1989-1990. AIDS 1992,6:495- 500. 107. Mortimer JY, Salathiel JA. 'Soundex' codes of surnames provide confidentiality and accuracy in a national HIV database. Commun Dis Rep CDR Rev I995,5:R183-6. 108. Swart AM, Weller I, Darby shire JH. Early HIV infection: to treat or not to treat? 5 M /1990,301:825-6. 109. Delta Co-ordinating Committee. Delta: a randomised double-blind controlled trial comparing combinations of zidovudine plus didanosine with zidovudine alone in HIV-infected individuals. Lancet 1996;348:283-91.
197 110. Alpha International Co-ordinating Committee. The Alpha trial: European/ Australian randomised double blind trial of two doses of didanosine in zidovudine intolerant patients with symptomatic HIV disease. AIDS 1996,:867-80. 111. Mulder JW, Cooper DA, Mathiesen L, et al. Zidovudine twice daily in asymptomatic subjects with HIV infection and a high risk of progression to AIDS: a randomised, double blind placebo controlled study. The European Australian Collaborative Group (Study 017). AIDS 1994,8:313-21. 112. Mannucci PM, Gringeri A, Savidge G, et al. Randomised double blind, placebo controlled trial of twice daily zidovudine in asymptomatic haemophiliacs infected with the human immunodeficiency vims type 1. European Australian Haemophilia Collaborative Study Group. Br J Haematol 1994,86:174-9. 113. Gibb DM, Darby shire JH, Debre M, Giaquinto C, Aboulker JP, Martinez M, Tudor Williams G. Treatment of children with HIV infection. PENT A (Paediatric European Network for Treatment of AIDS).Lancet 1995,345:1115. 114. CD4 Collaborative Group. CD4 surveillance in Scotland: perspectives on severe HIV-related immunodeficiency. AIDS 1997;11:1509-17 . 115. Moleswortb AM. Results from the 1995 survey of prevalent clinically diagnosed HIV infections in England, Wales, and Northern Ireland. Comm Dis Rep 1997;7(6):R77-82. 116. Norton J, Elford J, Sberr L, Miller R, Johnson MA. Repeat HIV testers at a London same-day testing clinic. AIDS 1997;11:773-81. 117. Pedersen C, Lindbardt B0, Jensen BL, et al. Clinical course of primary HIV infection: consequences for subsequent course of infection.BMJ 1989;299:154-7. 118. Lindback S, Brostrom, Karlsson A, Gaines H. Does symptomatic primary HIV-1 infection accelerate progression to CDC stage IV disease, CD4 count below
2 0 0 x l0 V l, AIDS, and death from AIDS?BMJ 1994;309:1535-7. 119. Dorrucci M, Rezza G, Vlahov D, et al. Clinical characteristics and prognostic value of acute retroviral syndrome among injecting drug users. Italian Seroconversion Study. AIDS 1995;9:597-604. 120. Byers RH Jr, Morgan WM, Darrow WW, et al. Estimating AIDS infection rates in the San Francisco Cohort.AIDS 1988;2:207-10. 121. Clayton C, Hills M. Statistical Models in Epidemiology. Oxford University Press, Oxford 1993.
198 122. Evans BG. The extent of under-reporting of AIDS cases in England, Wales and Northern Ireland. June 1995 DM thesis. 123. Crowe SM, Carlin JB, Stewart Kl, Luca CR, Hoy JF. Predictive value of CD4 lymphocyte numbers for the development of opportunistic infections and malignancies in HIV-infected persons. J AIDS 1991;4:770-776. 124. Beral V, Peterman V, Berkelman RL, Jaffe HW. Kaposi's sarcoma among persons with AIDS: a sexually transmitted infection? Lancet 1990;335:123-8. 125. ChangY, Moore PS. Kaposi's Sarcoma (KS)-associated herpesvirus and its role in KS. Infect Agents Dis 1996;5:215-22. 126. Kaplan EL, Meier P. Non-parametric estimation from incomplete observations.J Am Stat Assoc 1958;53:457-81. 127. Cox DR, Oakes D. Analysis of survival data. London: Chapman and Hall, 1984. 128. Cox DR. Regression models and life tables (with discussion). J Roy Stat Soc B 1972;34:187-220. 129. StataCorp. 1997. Stata Statistical Software: Release 5.0 College Station, TX: Stata Corporation. 130. Spijkerman IJB, Langendam MW, Veugelers PJ, et al. Differences in progression to AIDS between injection drug users and homosexual men with documented dates of seroconversion.Epidemiology 1996;7:571-7. 131. Prins M, Veugelers PJ for the European Seroconverter Study and the Tricontinental Seroconverter Study. Comparison of progression and non progression in injecting drug users and homosexual men with documented dates of HIV-1 seroconversion. 1997;11:621-31. 132. Eskild A, Magnus P, Brekke T, et al. The impact of exposure group on the progression rate to Acquired Immunodeficiency Syndrome. Scand J Infect Dis 1997;29:103-9. 133. Pezzotti P, Phillips AN, Dorrucci M,et al. Category of exposure to HIV and age in the progression to AIDS: longitudinal study of 1199 people with known dates of seroconversion. 5M 71996;313:583-6. 134. Munoz A, Xu J. Models for the incubation of AIDS and variations according to age and period. Stat Med 1996;15:2459-73. 135. Brettle RP, McNeil AJ, Bums S, et al. Progression of HIV: follow up of Edinburgh injecting dmg users with narrow seroconversion intervals in 1983- 1985. AIDS 1996;10:419-30.
199 136. van Benthem BHB, Veugelers PJ, Schechter MT, Kaldor JM, Page-Shafer KA, van Griensven GJP. Modelling the AIDS incubation time: evaluation of the right censoring strategies. AIDS 1997;11:834-5. 137. Sinicco A, Fora R, Raiteri R, Sciandra M, Bechis G, Calvo MM, Gioannini P. Is the clinical course of HIV-1 changing? Cohort Study. BMJ 1997;314:1232-7. 138. Keet IPM, Veugelers PJ, Koot M,et al. Temporal trends of the natural history of HIV-1 infection following seroconversion between 1984 and 1993. AIDS 1996;10:1601-2. 139. Volberding PA, Lagakos SW, Grimes JM, Stein DS, Balfour HH, Reichman RC, et al. The duration of zidovudine benefit in persons with asymptomatic HIV infection. Prolonged evaluation of protocol 019 of the AIDS Clinical Trials Group. 1994;272:437-42. 140. Lifson AR, Hessol NA, Buchbinder SP, Holmberg SD. The association of clinical conditions and serologic tests with CD4+ lymphocyte counts in HIV infected subjects without AIDS. AIDS 1991,5:1209-15. 141. Phillips AN, Lee CA, Elford J, Janossy G, Timms A, Bofill M, Kemoff PB. Serial CD4 lymphocyte counts and development of AIDS. Lancet 1991,337:389-92. 142. Phillips AN. CD4 lymphocyte depletion prior to the development of AIDS.AIDS 1992,6:735-6. 143. Phillips AN, Lee CA, Elford J, Janossy G, Kemoff PB. The cumulative risk of AIDS as the CD4 lymphocyte count declines. J Acquir Immune Defic Syndr 1992,5:148-52. 144. Mihailov C, Lamour A, Beaudre Bellein V, et al. Prognostic significance of cytotoxic T cells in individuals infected with human immunodeficiency vims. J Clin Immunol 1993,13:139-44. 145. Phillips AN, Sabin CA, Elford J, Bofill M, Janossy G, Lee CA. Acquired immunodeficiency syndrome (AIDS) risk in recent and long standing human immunodeficiency vims type 1 (HIV 1) infected patients with similar CD4 lymphocyte counts. Am J Epidemiol 1993,138:870-8. 146. Phillips AN, Pezzotti P, Cozzi Lepri A, Rezza G. CD4 lymphocyte count as a determinant of the time from HIV seroconversion to AIDS and death from AIDS: evidence from the Italian Seroconversion Study. AIDS 1994,8:1299-305. 147. Multicohort Analysis Project Workshop Part I. Immunologic markers of AIDS progression: consistency across five HIV infected cohorts.AIDS 1994,8:911-21.
200 148. Phillips AN, Sabin CA, Elford J, Bofill M, Janossy G, Lee CA. Use of CD4 lymphocyte count to predict long term survival free of AIDS after HIV infection. 5 M 7 1994,309:309-13. 149. Sabin CA, Mocroft A, Phillips AN. The use of CD4 counts as prognostic markers in HIV infection. AIDS 1995,9:1205-6. 150. Vanhems P, Allard R, Toma E, Cyr L, Beaulieu R. Prognostic value of the CD4+ T cell count for HIV 1 infected patients with advanced immunosuppression. Int J STD AIDS 1996,7:495-501. 151. Saravolatz L, Neaton JD, Sacks L, Deyton L, Rhame F, Sherer R. CD4+ T lymphocyte counts and patterns of mortality among patients infected with human immunodeficiency virus who were enrolled in community programs for clinical research on AIDS. Clin Infect Dis 1996,22:513-20. 152. Mellors JW, Muftoz A, Giorgi JV, et al. Plasma viral load and CD4+ lymphocytes as prognostic markers of HIV 1 infection.Ann Intern Med 1997,126:946-54.
201