HIV Indicator Diseases Across Europe
Statistical Analysis Plan
May 2010
1. Objectives
The objectives of the indicator diseases study are to launch, implement and evaluate a survey initiative to assess HIV prevalence for one or more diseases and/or conditions within a specific segment of the population not yet diagnosed with HIV and that present for care with the specific disease/condition.
The implementation will be conducted in two phases. Phase one will be to implement the survey based on a list of 8 diseases associated with high-risk behaviour or immune deficiency (listed below).
List of phase one Indicator Diseases
(1) Presenting for care of a sexually transmitted disease (including gonorrhoea, syphilis and other ulcerative genital conditions and chlamydia),
(2) Presenting for care of malignant lymphoma, irrespective of type
(3) Presenting for care of cervical or anal dysplasia or cancer,
(4) Presenting for care of herpes zoster in a person younger than 65 years,
(5) Hepatitis B or C virus infection (acute or chronic – and irrespective of time of diagnosis relative to time of survey),
(6) Presenting with ongoing mononucleosis-like illness
(7) Presenting with unexplained leukocytopenia or thrombocytopenia lasting at least 4 weeks
(8) Presenting with seborrheic dermatitis / exanthema
An evaluation will take place following phase one with changes implemented for initiating phase two. Phase two will involve opening up for a wider list of diseases and analysis.
2. Included Patients
Patients are included with the following inclusion criteria;
(1) Presenting for care of a sexually transmitted disease (including gonorrhoea, syphilis and other ulcerative genital conditions and chlamydia),
(2) Presenting for care of malignant lymphoma, irrespective of type
(3) Presenting for care of cervical or anal dysplasia or cancer,
(4) Presenting for care of herpes zoster in a person younger than 65 years,
(5) Hepatitis B or C virus infection (acute or chronic – and irrespective of time of diagnosis relative to time of survey),
(6) Presenting with ongoing mononucleosis-like illness
(7) Presenting with unexplained leukocytopenia or thrombocytopenia lasting at least 4 weeks
(8) Presenting with seborrheic dermatitis / exanthema
3-5 surveys are performed per disease of a minimum of 100 patients per survey.
One survey assesses HIV prevalence for one specific disease/condition for a specific segment of the population within a specific setting.
The survey will be implemented on consecutive patients not yet known to be HIV-infected and that present with only one of the 8 conditions mentioned above from a given day until a pre-specified number of persons have entered the survey (at least 100 patients but preferentially 200-400 being tested for HIV).
3. Statistical Methods
The recruitment process will be described, in terms of a qualitative description of the recruitment process, barriers to recruitment and issues encountered which have affected recruitment to Phase I of the study and which may potentially influence Phase II of the study. The rate of recruitment over time will be described (potentially stratified by centre, region or indicator disease). Generalisability of the patients included will be considered by a qualitative description of the patients who were not included in the study, potential reasons for this and variations between centres and indicator diseases (note that this has not been recorded for all surveys).
Descriptive analyses will be used to summarise the tested patients, using information available on Form A and B, including
· Age
· Gender
· Sexuality
· Previous test history etc
and
· Percentage of patients previously tested for HIV
· Percentage of patients (not) returned for test results
· Percentage of patients tested where testing is not routine procedure for the specific indicator disease
Prevalence estimates of HIV infection (with 95% confidence intervals) will show the following:
· HIV prevalence per disease/condition per specific setting (note that some diseases/conditions have sub-categories, STDs, Cervical and anal dysplasia and cancer, Hepatitis B or C)
· HIV prevalence per disease
· HIV prevalence per disease per region (if possible)
Logistic regression analysis will be used to identify the factors associated with, for example, testing HIV-positive. Detailed analyses will depend on the number of patients enrolled and testing positive; it may be necessary to combine diseases and regions in order to provide meaningful results. For example, a decision could be made a priori to analyse each indicator disease where >30 patients have tested HIV positive, or alternatively to combine some indicator diseases. Adjustments will be made for collected variables, such as type of disease tested for, regional setting, demographic information, previous testing history etc.
For those patients testing positive, analysis of previous symptoms, progression of disease at diagnosis (number of late presentation and presentation with advanced HIV disease). If there were sufficient numbers of patients testing positive, logistic regression could be used to identify the factors associated with late presentation, to identify possible areas for intervention on late presentation. Similarly, depending on numbers, logistic regression could be used to further explore which factors are associated with returning for the test result; patients with a negative and positive test could be compared or for those testing positive but not returning, the characteristics could be compared to those who do return in order to identify patients least likely to return for possible targeted intervention.
The focus on indicator diseases to increase testing will need a country – or region - specific analysis of where people are not tested as well as an analysis of what this tool will mean in terms of getting people tested earlier.
4