APPENDIX A

EXPOSURE CATEGORY ADJUSTMENTS

We developed and applied a procedure to adjust for unknown region, unknown sex, and unknown exposure category among first-time HIV-positive diagnoses, 1985 to 2010 Adjustments were completed in five main steps; similar steps were carried out for each modified health region and then summed to obtain Ontario totals. Calculations were carried out in Microsoft Excel Fractional cases were retained through all five steps and rounded only in the final tables.

Step 1: Distribute HIV diagnoses among males, females, unknown sex with unknown health region among the males, females, unknown sex in the seven health regions in accordance with the proportion among the known.

i) Obtain the distribution of HIV-positive results for each health region, including unknown region, by sex for each year and exposure category

ii) Assign HIV-positive results in males, females, unknown sex in unknown region to the seven health regions in accordance with the distribution among the known.

Example:
In 1991, unknown region, exposure category NIR, there were 180 diagnoses in males, 17 in females and 18 with sex unknown. That same year in Toronto, exposure category NIR, 538 cases were among males, 57 among females and 84 unknown sex. Provincial totals for 1991, exposure category NIR, were 1,049 diagnoses among males, 121 among females and 122 unknown sex. To allocate the number of cases by sex with unknown region to Toronto, the formula used was as follows:

# Toronto, NIR + # unknown region, NIR x [# Toronto, NIR / (# Ontario NIR - # unknown region, NIR)]

For males, the calculation was:

538 + 180 x [ 538 / (1,049 - 180) ] = 649.4

which was the ‘adjusted' number of HIV-positive results among Toronto males in the exposure category NIR in 1991.

Similarly, the adjusted positive results among females, Toronto, exposure category NIR was:

57 + 17 x [ 57 / (121 - 17) ] = 66.3

and for unknown sex; the adjusted number of HIV-positive results among unknown sex, Toronto, NIR in 1991 was:

84 + 18 x [ 84 / (122 - 18) ] = 98.5


This procedure was repeated by sex (i.e. for males, females, and unknown sex), year (1985, 1986, etc. to 2010) and exposure category (MSM, MSM-IDU, etc. Other, NIR) and, in this manner, HIV-positive diagnoses with unknown health region were distributed among the seven health regions. Subsequent steps were completed within each of the seven modified health regions.

Go to Top

Step 2: Assign HIV diagnoses with unknown sex to males and females in accordance with the proportion among those with known sex.

After allocating HIV-positive results in unknown health region among males, females and unknown sex in each of the seven regions (Step 1), HIV-positive results with unknown sex within each region were allocated to males or females within that region.

Example:

In 1991 in Toronto, there were 98.5 HIV-positive results with unknown sex in exposure category NIR (calculated in Step 1). These were allocated to the adjusted number of males or females in 1991, exposure NIR which had already been adjusted for unknown region. For Toronto men, we used the following formula:

# males + # unknown sex x [ # males / (# males + # females) ]

Therefore, the number of HIV-positive results among Toronto males in 1991, exposure NIR, adjusted for unknown sex was:

649.4 + 98.5 x [ 649.4 / (649.4 + 66.3) ] = 738.8

and among females:

66.3 + 98.5 x [ 66.3 / (649.4 + 66.3) ]= 75.4

In this manner, the total number of HIV-positive results in Toronto in 1991, exposure category NIR, that is, 649.4 males + 66.3 females + 98.5 unknown sex = 814.2 were adjusted to 738.8 males + 75.4 females = 814.2 HIV positive results. This procedure was repeated for each year, each exposure category and each of the seven health regions.

Go to Top

Step 3: Reallocate HIV diagnoses in each exposure category according to new distribution by the Laboratory Enhancement  Study (LES) from 1999 to 2007.

Step 3.1 For each exposure category and sex, calculate the LES adjustment factors. Regions for which reallocation for MSM to MSM-IDU or IDU to MSM-IDU exposure categories were aggregated in homogeneous geographic groups (not necessarily contiguous) For other exposure categories, each region had its own adjustment factors by gender and also took into account trends in the distribution of cases from 1999 to 2007 in the LES. For HIV-negative results, only cases in the low risk heterosexual exposure category were generally reallocated according to the distribution by gender in each region.

 

Step 3.2 For each sex, each exposure category and each year from 1985 to 2010, calculated the number of cases that was going to be subtracted from that exposure category.

Example:

Among Toronto males in 1985, there were 114.1 HIV-positive results in the MSM category (calculated in Step 2). The LES adjustment factor for the MSM category for that region was 1.3%. Therefore, the number of cases that was reallocated from that category was:

114.1 *1.3% = 1.48 cases

Step 3.3 For each sex, each exposure category and each year, calculated the number of cases that was going to be reallocated to that exposure category.

Example:

Among Toronto males in 1985 ,there were 114.1 cases in MSM, 3.0 in MSM-IDU and 107.2 in NIR (Step 2). In Step 3.1, we calculated that only 1.3% of MSM cases in Step 2 were reallocated to the MSM-IDU category. Therefore, the number of cases that was reallocated to the MSM-IDU category was:

(114.1* 1.3%)+(3.0* 0%)+ (107.2 * 0%) = 1.48 cases

Step 3.4 For each sex, each exposure category and each year, calculate the final reallocated number of cases.

Example:

The MSM category in Toronto males in 1985 had 114.1 cases (step 2), 1.48 cases were reallocated to another category (Step3.2) and none was reallocated to MSM itself (Step 3.3). Therefore, the total number after reallocation was:

114.1 - 1.48 + 0 = 112.6 cases

Go to Top

Step 4: Allocate HIV-positive results among exposure category NIR to known exposure categories.

Step 4.1 For each exposure category, for each sex (males, females) within each year, calculate the proportion of HIV-positive results which had that exposure that year.

Example:

Among Toronto males in 1991, there were 1.4 HIV-positive results with exposure LR, 738.8 positive results with exposure NIR (calculated in Step 3) and a total of 1,119.0 positive results that year. Therefore, the proportion of HIV-positive results in exposure LR hetero was:

1.4 / (1,119.0 - 738.8) x 100% = 0.38%

For Toronto females in 1991, there were 1.7 positive results with exposure LR hetero, 75.4 positive results with exposure NIR (Step 3) and a total of 88.0 positive results that year. The proportion of positives in exposure LR hetero was:

1.7 / (88.0 - 75.4) x 100% = 15.9%

Step 4.2 For each exposure category, for each sex and each region, list the Lab enhancement study (LES) adjustment factors, which were token into account trends of distribution from 1999 to 2007 by LES. For example, according to the trend of distribution from 1999 to 2007 in the LES, HIV-positive male cases in Toronto were grouped into two periods: from 1999 to 2005 and from 2006 to 2007. LES adjustment factors were 0.0% for exposures of Clotting factor and MTC.

Step 4.3 For each exposure, each sex, for the years 1999 and 2010 only, calculate the average of the proportion among the known (Step 4.1).

Example:

In Toronto males in 1999, the proportion of HIV-positive results with exposure MSM was 78.2% and in 2000, was 78.7%, 74.7% in 2001, 76.2% in 2002, 78.3% in 2003, 82.0% in 2004, and 79.5% in 2005, giving an average proportion for the seven years of 78.3%.

Step 4.4 For each year for each sex and each exposure category, calculate the “scaled-back” proportion of HIV-positive results in that exposure category that year using the formula:

proportion among the known x (LES adjustment factor / average proportion in 1999-2010)

Component 1 of the formula took into account the fact that the proportion of HIV-positive results by exposure category had shifted over time, for example, early in the epidemic, most HIV-positive results were in the exposure category of MSM but new diagnoses in this group had declined over time.

Component 2 took into account the inappropriateness of applying in isolation the LES adjustment factors, based on data collected in 1999 and 2007, to HIV-positive results diagnosed 10 to 15 years earlier. Component 3 of the formula incorporated data on HIV-positive results which might or might not have contributed to the LES adjustment factors (study questionnaire was not returned).

Example:

In Toronto males in 1985, the proportion of HIV-positive results among MSM was 96.2% (proportion among the known as calculated in Step 4.1), the LES adjustment factor for Toronto males, MSM was 56.9% (Step 4.2) and the average proportion among the known for 1999 to 2005 was 78.3% (Step 4.3). Using the formula in Step 4.4, the scaled-back adjustment factor for 1985 was:

96.2% x (56.9% / 78.3%) = 70.0%

This step was repeated for each year, each sex and each exposure category. In the event that the LES adjustment factor was 0.0%, we used the proportion among the known, unless the exposure category was Clotting factor or MTC, in which cases the adjustment factor remained 0.0% (no HIV-positive results from NIR were to be assigned to these two categories).

Step 4.5 The scaled-back adjustment factors for each exposure category within each year were then standardized to sum to 1.0 since the sum of the proportions calculated in Step 4.4 in each exposure category in each year did not necessarily add to 100%.

Example:

In 1985 in Toronto, the sum of the scaled-back proportions calculated in Step 4.4 for males was 73.0%. The proportions in each exposure category were "normalized to 1.0" by dividing the proportion in that exposure category by the sum of the proportions that year. For MSM in Toronto males that year, the calculation was:

70.0% / 73.0% = 95.8%

The process was repeated for each exposure category for each sex for each year and in this manner, final adjustment factors were generated for the health region.

Go to Top

Step 5: Calculate the final number of diagnoses, adjusted for unknown region, sex, known and unknown exposure, for each year for each sex in each exposure category.

To calculate the adjusted number of diagnoses for males or females for a given exposure category in a given year, the final adjustment factor calculated in Step 4.5 was multiplied by the number of HIV-positive with unknown exposure that year and added to the HIV-positive tests with known exposure.

Example:

In Toronto males in 1985, exposure category MSM, the final adjustment factor was 95.8% (Step 4.5), there were 112.6 HIV-positive results among MSM that year (adjusted for unknown region, unknown sex and reallocated exposure category) and 107.2 HIV-positive results in exposure NIR. Therefore, the adjusted number of HIV-positive results among Toronto males in 1985 was:

112.6 + 95.8% x 107.2 = 215.4 HIV-positive results

This calculation was repeated for each exposure category for each year for HIV-positive results among males and females. Ontario totals for each sex by year and exposure category (as seen in Table 1.5), were obtained by summation across the regions.

The same methodology was used to assign HIV-negative results of unknown region, unknown sex and unknown exposure category for each year 1992 to 2010 to the seven health regions. Regionally adjusted HIV-negative tests per exposure category were summed to provide provincial totals. HIV positivity rates for each modified health region by year of diagnosis (1992, 1993, etc., 2010) and exposure category were calculated using adjusted figures such that the number of HIV tests (adjusted) was the sum of HIV-positive results + HIV-negative results adjusted as described above.

Last Updated: May 04, 2011