Older HIV-Infected Individuals: Presentation and Mortality
Older HIV-Infected Individuals: Presentation and Mortality
2038 patients aged over 18 years attended the sexual health clinic in Brighton between 1 January 1996 and 31 May 2010. Excluded from the analyses for late presentation were 493 (24.2%) patients with no documented baseline CD4 cell count for reasons stated in Figure 1A and a further 9 patients who had no documented date of HIV diagnosis, leaving 1536 (75.4%) patients with a documented CD4 cell count within 3 months of diagnosis eligible for this analysis.
(Enlarge Image)
Figure 1.
Cohort construction.
For the analysis on mortality, 64 (3.5%) patients only attended once and did not contribute to follow-up time, while the dates of HIV diagnosis were missing for 9 patients leaving 1965 (96.4%) patients eligible for this analysis (Figure 1B). Ethical approval was not sought for this study as we utilized non-identifiable information previously collected in the course of normal care. Written informed consent was not required for the same reason, however access to the clinic database was approved by the HIV/Genitourinary Medicine Protocol Review Panel at Brighton and Sussex University Hospitals NHS Trust.
Patients testing HIV positive that were either diagnosed within the clinic or referred from another facilty were followed up at 3 monthly intervals. Demographic information collected at first visit were age at HIV diagnosis, date of HIV diagnosis, ethnicity and HIV risk group. The clinic database also had information on whether patients had been previously diagnosed elsewhere with HIV or not. This information ensured that true baseline CD4 cell counts were used for the analyses presented.
Patients initiating ART attended at 2, 6 and 10 weeks for blood tests and saw a doctor at 4, 8 and 12 weeks and 3 monthly subsequently after full suppression of the HIV RNA viral load.
CD4 cell count and HIV viral load within 90 days of HIV diagnosis were documented and subsequently measured at 3 monthly intervals in stable patients or more frequently at the discretion of their physician. The causes of death were entered into the clinic database from death certificates. Causes of death were ascertained for patients who died outside a healthcare facility following a coroner post-mortem examination and this was also entered into the clinic database when the information became available.
To examine the proportion that presented late by calendar time stratified by age, calendar period of diagnosis was categorized as 1996–2001, 2002–2005 and 2006–2010 so that there were adequate numbers in each group for statistical analyses. Mean and standard deviations were reported for normally distributed variables, and median and inter-quartile ranges were used for skewed distributions. Frequencies and percentages were reported for categorical exposure variables.
Factors associated with late presentation were assessed using odds ratios (ORs) estimated using logistic regression. The likelihood ratio test statistic was used to estimate the p-values for association in the multivariable logistic regression model.
We also examined for the median time to initiation of HAART for the two age groups stratified by CD4 cell count at presentation using the Kaplan-Meier approach. For this analysis, patients were censored at the time started HAART for those on treatment and time last seen for those who remained untreated at the end of follow-up.
To estimate mortality rates, participants were censored at the earliest of: death, date last seen, or end of follow-up (31 May 2010). Poisson regression modelling was used to estimate survival probability. Patients who attended at least once following the enrollment visit contributed person-years (pyr) for the analyses. Age was the primary exposure when fitting the multivariable Poisson regression model. This was adjusted for a priori confounders such as ethnicity and HIV risk group irrespective of whether they acted as confounders in this study or not. All analyses were carried out using STATA 11.0 (StataCorp, College Station, TX, USA).
Methods
Study Participants
2038 patients aged over 18 years attended the sexual health clinic in Brighton between 1 January 1996 and 31 May 2010. Excluded from the analyses for late presentation were 493 (24.2%) patients with no documented baseline CD4 cell count for reasons stated in Figure 1A and a further 9 patients who had no documented date of HIV diagnosis, leaving 1536 (75.4%) patients with a documented CD4 cell count within 3 months of diagnosis eligible for this analysis.
(Enlarge Image)
Figure 1.
Cohort construction.
For the analysis on mortality, 64 (3.5%) patients only attended once and did not contribute to follow-up time, while the dates of HIV diagnosis were missing for 9 patients leaving 1965 (96.4%) patients eligible for this analysis (Figure 1B). Ethical approval was not sought for this study as we utilized non-identifiable information previously collected in the course of normal care. Written informed consent was not required for the same reason, however access to the clinic database was approved by the HIV/Genitourinary Medicine Protocol Review Panel at Brighton and Sussex University Hospitals NHS Trust.
Study/Clinic Procedures
Patients testing HIV positive that were either diagnosed within the clinic or referred from another facilty were followed up at 3 monthly intervals. Demographic information collected at first visit were age at HIV diagnosis, date of HIV diagnosis, ethnicity and HIV risk group. The clinic database also had information on whether patients had been previously diagnosed elsewhere with HIV or not. This information ensured that true baseline CD4 cell counts were used for the analyses presented.
Patients initiating ART attended at 2, 6 and 10 weeks for blood tests and saw a doctor at 4, 8 and 12 weeks and 3 monthly subsequently after full suppression of the HIV RNA viral load.
CD4 cell count and HIV viral load within 90 days of HIV diagnosis were documented and subsequently measured at 3 monthly intervals in stable patients or more frequently at the discretion of their physician. The causes of death were entered into the clinic database from death certificates. Causes of death were ascertained for patients who died outside a healthcare facility following a coroner post-mortem examination and this was also entered into the clinic database when the information became available.
Statistical Analyses
To examine the proportion that presented late by calendar time stratified by age, calendar period of diagnosis was categorized as 1996–2001, 2002–2005 and 2006–2010 so that there were adequate numbers in each group for statistical analyses. Mean and standard deviations were reported for normally distributed variables, and median and inter-quartile ranges were used for skewed distributions. Frequencies and percentages were reported for categorical exposure variables.
Factors associated with late presentation were assessed using odds ratios (ORs) estimated using logistic regression. The likelihood ratio test statistic was used to estimate the p-values for association in the multivariable logistic regression model.
We also examined for the median time to initiation of HAART for the two age groups stratified by CD4 cell count at presentation using the Kaplan-Meier approach. For this analysis, patients were censored at the time started HAART for those on treatment and time last seen for those who remained untreated at the end of follow-up.
To estimate mortality rates, participants were censored at the earliest of: death, date last seen, or end of follow-up (31 May 2010). Poisson regression modelling was used to estimate survival probability. Patients who attended at least once following the enrollment visit contributed person-years (pyr) for the analyses. Age was the primary exposure when fitting the multivariable Poisson regression model. This was adjusted for a priori confounders such as ethnicity and HIV risk group irrespective of whether they acted as confounders in this study or not. All analyses were carried out using STATA 11.0 (StataCorp, College Station, TX, USA).