• No results found

Trends over time in prevalence of virological non-suppression

VL among individuals currently receiving continuous cART for at least six months

This section focuses on trends in virological non-suppression on cART. This first sub- section considers the sub-population described in Section 5.3.3, and focuses on individuals currently receiving continuous cART for at least 6 months at the time of the VL measurement. The second column of Table 5.1 shows the denominators used for these analyses.

As shown in Figure 5.11, the prevalence of virological non-suppression among individuals currently receiving continuous cART fell dramatically between 2000 and 2014. For all gender/sexual orientation groups, the decline appeared to occur predominantly between 2000 and 2006, with only a marginal decline subsequently. Among MSM, the prevalence of virological non-suppression fell from 31% in 2000 to just 8% in 2014. Likewise, among MSW and women, 39% and 31% had virological non-suppression in 2000, respectively, compared to 11% of each group in 2014. A greater percentage of MSW had virological non-suppression than MSM in all calendar years and a greater percentage of women had virological non-suppression than MSM from 2005 onward. Examining the figures visually, the pattern of the reduction in the prevalence of virological non-suppression over time (shown by the slope of the line) appeared similar in all gender/sexual orientation groups.

185

Figure 5.11: Prevalence of VL >50 copies/mL by gender/sexual orientation among individuals currently receiving continuous cART for at least six months 2000-14 a b 0 5 10 15 20 25 30 35 40 45 50 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 Pr ev al en ce o f V L >5 0 co pi es /m L (% ) Calendar year MSM MSW Women

a Individuals could be included at more than one time point; b denominators are provided in column two of Table 5.1; the bars on graph represent confidence intervals.

186 The results presented in Figure 5.11 suggested a non-linear association between calendar time and virological non-suppression. Furthermore, when including a quadratic term for calendar year in the model, this was strongly associated with virological non-suppression (p<0.0001 in a model including calendar year and calendar year squared only) which provided further evidence of a non-linear effect. Thus, as described in the methods section, I fitted a piecewise linear slope to investigate changes over calendar time in the prevalence of virological non-

suppression. Using Figure 5.11, I decided to allow for a change in the slope at the year 2006. The results of the piecewise model demonstrated strong evidence for a change in slope in 2006: in 2006-14 the change in prevalence per year was 0.84 times that seen in 2000-06 (95% CI: 0.81, 0.87; p<0.0001). This corresponded to an 18% relative reduction in prevalence of virological non-suppression per year before 2006, and a 2% relative reduction in prevalence of virological non-suppression per year after 2006, in a model including only gender/sexual orientation and the two calendar year terms (Table 5.3). After adjustment for age and new patient status, there remained a strong

downward trend in the prevalence of virological non-suppression per calendar year before 2006; however, after 2006 there was no evidence of a decline over time. Table 5.3 also shows the associations between gender/sexual orientation and virological non-suppression. There were significant differences between each of the gender/sexual orientation groups. The prevalence ratio of 1.56 comparing MSW and MSM means that MSW had a 56% greater prevalence of virological non-suppression compared to MSM across the whole time-period. Similarly, women had 24% greater prevalence of virological non-suppression compared to MSM. In addition, there was evidence of a 19% lower prevalence of virological non-suppression among women compared to MSW. Following additional adjustment for age and new patient status, there remained evidence of differences between in prevalence of non-suppression between each of the gender/sexual orientation groups, with MSM having the most favourable profile and MSW the least favourable. Younger age and less than six months since an individual’s first visit to the ICDC compared to >12 months were associated with a substantially higher prevalence of virological non-suppression.

187

Table 5.3: Association between gender/sexual orientation, calendar year and VL >50 copies/mL among individuals currently receiving continuous cART for at least six months ab

Test for interaction between gender/sexual orientation and calendar year In all models fitted until now, I have assumed that all three gender/sexual orientation groups had the same change over calendar time in terms of the relative prevalence of virological non-suppression. This section investigates whether there were different slopes for each gender/sexual orientation group. In other words, I am assessed

whether there were different relative changes over time in three groups, and therefore, whether or not the MSW and women were “catching up” or “falling behind” the MSM, by experiencing a faster or slower decline over time in the prevalence of virological non-suppression. This was achieved by a test for interaction between gender/sexual orientation and calendar year. As a result of the findings in the previous paragraph, calendar year was considered in the two periods: 2000-06 and 2006-14. The results are shown in Table 5.4. Recall that here I am considering the relative changes over time in virological non-suppression among the three groups. In the period 2000-06, MSM had an adjusted PR of 0.82, corresponding to an 18% reduction in the

prevalence of virological non-suppression per year later. Among MSW and women, these figures were 15% and 13%, respectively. Thus there were greater relative improvements over time for MSM than the other two groups before 2006, but the

Covariates Model including

gender/sexual orientation and calendar year

Model additionally including age and new

patient status aPR 95% CI P-valuec aPR 95% CI P-valuec Gender/ sexual orientation MSW vs. MSM 1.56 1.34, 1.81 <0.0001 1.59 1.37, 1.84 <0.0001 Women vs. MSM 1.24 1.09, 1.44 1.18 1.03, 1.35 Women vs. MSW 0.81 0.68, 0.96 0.74 0.63, 0.88 Calendar year 2000- 06 Per year 0.82 0.80, 0.84 <0.0001 0.84 0.82, 0.86 <0.0001 Calendar year 2006- 14 Per year 0.98 0.96, 1.00 0.045 1.00 0.98, 1.02 0.96

Age Per 10 years - - - 0.81 0.76, 0.86 <0.0001

New patient status (time since first visit to the ICDC) ≤6 months vs. >12 months - - - 2.22 1.97, 2.50 <0.0001 6-12 months vs. >12 months - - 1.12 0.95, 1.32

a PRs compare the relative positions of the slopes for VL >50 copies/mL over time, a slope which is assumed to stay the same over the study period; b individuals could be included at more than one time point; c likelihood ratio test; aPR= adjusted Prevalence Ratio.

188 differences between the groups were not statistically significant (p=0.15; test for

interaction). In contrast, there was no evidence of a change in the prevalence of virological non-suppression over time among MSM, MSW, or women between 2006 and 2014, nor that these changes over time were different between the three groups (p=0.58; test for interaction). In a model solely including the post-2006 period, similarly there was no evidence of differences between the gender/sexual orientation groups (p=0.62).

Table 5.4: Associations of the interaction between gender/sexual orientation and calendar year with virological non-suppression among individuals currently receiving continuous cART for at least six months a

VL among individuals who ever started cART

This section considers the second sub-population described in Section 5.3.3of those that have ever started cART more than six months previously, regardless of whether they interrupted or currently remained on cART. The final column of Table 5.1 shows the denominators for the analyses in this section.

When considering this sub-population, as opposed to those on continuous ART for at least six months (Section 5.4.2.1 ), the prevalence of VL >50 copies/mL was between 1.5% and 7% higher in each year. The difference in prevalence of virological non- suppression found between these sub-populations decreased over the study period. MSW and MSM have a comparable prevalence of virological non-suppression in 2006 (13% vs. 12% respectively), however, after this date the difference between the groups increased, with MSM having a lower prevalence (Figure 5.12). Women had an initially higher prevalence of virological non-suppression than MSM in 2006 (19% vs. 12% respectively), and this gap does not appear to have lessened over time, with substantial differences between these groups found in most years. MSW and women had a similar prevalence of virological non-suppression from 2008 onward, and in 2014, 13% of both groups had a VL >50 copies/mL compared to 9% of MSM.

aPR b 95% CI P-value for interaction c Relative change per year later:

2000-06

MSM 0.82 0.80, 0.85 0.15 MSW 0.85 0.80, 0.89

Women 0.87 0.83, 0.92 Relative change per year later:

2006-14

MSM 0.99 0.96, 1.02 0.58 MSW 1.01 0.97, 1.05

Women 1.01 0.97, 1.04

a Individuals could be included at more than one time point; b adjusted for age and new patient status; c likelihood ratio test examining whether there is a difference in change over time between the three gender/sexual orientation groups; aPR= adjusted Prevalence Ratio.

189

Figure 5.12: Prevalence of VL >50 copies/mL by gender/sexual orientation among individuals ever on cART 2006-14 a b

0 5 10 15 20 25 30 35 40 45 50 2006 2007 2008 2009 2010 2011 2012 2013 2014 P re v al en ce o f V L > 5 0 co p ie s /m L ( % ) Calendar year MSM MSW Women

a Individuals could be included at more than one time point; b denominators are provided in column three of Table 5.1; the bars on graph represent confidence intervals.

190 In this analysis, only considering 2006 onwards, there was no evidence of a non-linear trend between calendar year and VL non-suppression (p=0.72 in a model including calendar year and calendar year squared only), a finding which was corroborated by visual examination of Figure 5.12. Therefore, I considered calendar year as a

continuous linear variable in subsequent analyses.

Table 5.5 shows the PRs for the associations of gender/sexual orientation and calendar year with virological non-suppression when assuming the same relative change over time in prevalence of non-suppression in all three gender/sexual

orientation groups. In a model including gender/sexual orientation and calendar year only, there were higher prevalence’s of virological non-suppression among MSW and women compared to MSM (by 43% and 57% respectively), but no evidence of

differences between MSW and women. For each year later, there was a 6% lower prevalence of virological non-suppression on average. The effects of gender/sexual orientation and calendar year were marginally attenuated following additional adjustment for age and new patient status.

Table 5.5: Association between gender/sexual orientation, calendar year and VL >50 copies/mL among individuals ever on cART a b

Test for interaction between gender/sexual orientation and calendar year There was weak evidence of different changes over time in the prevalence of non- suppression between gender/sexual orientation groups when adjusting for age and new patient status (p=0.066; test for interaction) (Table 5.6). Both MSM and women

Covariates Model including

gender/sexual orientation and calendar year

Model additionally adjusted for age and new

patient status aPR 95% CI P-valuec aPR 95% CI P-valuec Gender/ sexual orientation MSW vs. MSM 1.43 1.20, 1.69 <0.0001 1.47 1.24, 1.73 <0.0001 Women vs. MSM 1.57 1.36, 1.81 1.41 1.23, 1.62 Women vs. MSW 1.10 0.92, 1.32 0.96 0.80, 1.15 Calendar year Per year 0.94 0.93, 0.96 <0.0001 0.96 0.95, 0.98 <0.0001 Age Per 10 years - - - 0.72 0.68, 0.77 <0.0001 New patient status (time since first visit to the ICDC) ≤6 months vs. >12 months - - - 1.71 1.47, 1.99 <0.0001 6-12 months vs. >12 months - - 1.03 0.85, 1.25

a PRs compare the relative positions of the slopes for VL >50 copies/mL over time, a slope which is assumed to stay the same over the study period; b individuals could be included at more than one time point; c likelihood ratio test; aPR= adjusted Prevalence Ratio.

191 had a 5% lower prevalence of virological non-suppression per year on average;

however, there was no evidence of a reduction in the prevalence of virological non- suppression over calendar time among MSW. This may reflect the fact that overall prevalence of virological non-suppression in this latter group was particularly low at the start of the period under consideration, and as such could not decrease substantially further.

Table 5.6: Associations of the interaction between gender/sexual orientation and calendar year with virological non-suppression among individuals ever on cART a