First Contact Details
- Participant Flow Diagram & Exclusions
- Socio-demographic Characteristics
- Substance Use Profile
- Quantitative Description of Tapentadol Use
- Injection Drug Use & Related Characteristics
- Stigma and Motivation to change
- HCV - Awareness, Prevalence & Treatment Continuum
- Treatment Experience & Outcome for Opioid Use Disorder
- Qualitative Description of Tapentadol Use
Participant Flow Diagram & Exclusions
Geographical Locations and Clustering
As defined in the inclusion criteria, participants were recruited from Bengaluru district. Two subjects who acquired the habit in Bangalore but then shifted the residence to other districts of the Karnataka have been retained in the analysis.
The figures below show the general location of all participants and that the biggest cluster of patients are from near NIMHANS.
Socio-demographic Characteristics
A total of 272 participants were recruited for this study. However,
two entries in community arm were later identified as duplicates where
the same patient gave different name. Only one of the entries are
retained leaving us with 270 unique participants.
As described in the methodology, we had marked cases as suspicious when
we were not able to ascertain if they were PWIDs one way or another. 8
cases are retained in the analysis. For HCV seroprevalence results we
report results with and without including them.
The participants belong to two distinct groups:
- Convenience Sample (recruited at NIMHANS).
- Community Sample (recruited via peer-referrals and linkages).
The following table shows the basic demographic details of the participants:
Characteristic | Community | Convenience | p-value2 |
|---|---|---|---|
Age | 26 (24, 30) | 24 (22, 26) | <0.001 |
Gender | >0.9 | ||
Male | 93 (100%) | 176 (99%) | |
Female | 0 (0%) | 1 (0.6%) | |
Years of education | 8 (7, 10) | 9 (7, 10) | 0.001 |
Occupation as classified in nfhs | 0.4 | ||
Skilled and unskilled manual | 65 (70%) | 109 (62%) | |
Not working | 9 (9.7%) | 26 (15%) | |
Sales | 9 (9.7%) | 24 (14%) | |
Services / household and domestic | 9 (9.7%) | 11 (6.2%) | |
Professional / technical / managerial | 1 (1.1%) | 6 (3.4%) | |
Clerical | 0 (0%) | 1 (0.6%) | |
Is a gig worker? | 23 (25%) | 54 (31%) | 0.3 |
Kuppuswamy socio-economic status class | 0.3 | ||
Upper middle class | 42 (45%) | 68 (38%) | |
Middle class | 35 (38%) | 71 (40%) | |
Lower middle class | 9 (9.7%) | 29 (16%) | |
Upper class | 6 (6.5%) | 9 (5.1%) | |
Lower class | 1 (1.1%) | 0 (0%) | |
1Median (Q1, Q3); n (%) | |||
2Wilcoxon rank sum test; Fisher's exact test; Pearson's Chi-squared test | |||
Gig worker as defined in the Karnataka Draft Bill | |||
Community recruited patients are on average 2 years older than convenience sample
We see a preponderance of manual work as occupation and approximately one in four patients are ‘Gig Workers’ i.e. people who do not have much social security through employment.
Amongst the patients recruited from the hospital (convenience sample), most came to know about treatment services via thier peers (Figure below).
Detailed Description of Socio-economic status
Although, the Kuppuswamy scale shows most patients to be in upper middle socio-economic status we did a finer analysis with reported monthly income and number of members in the household.
Characteristic | Community | Convenience | p-value2 |
|---|---|---|---|
Self reported monthly earning of the household | 9680 (7090, 12550) | 7950 (7060, 12160) | 0.4 |
What is the total number of people in the household? | 4 (3, 5) | 4 (3, 5) | 0.013 |
Calculated monthly per capita income | 2337 (1800, 3517) | 2033 (1512, 3205) | 0.047 |
Education of head of the family | 0.046 | ||
Middle school certificate | 26 (28%) | 53 (30%) | |
High school certificate | 21 (23%) | 44 (25%) | |
Illiterate | 23 (25%) | 23 (13%) | |
Primary school certificate | 8 (8.6%) | 35 (20%) | |
Intermediate or post-high-school diploma | 9 (9.7%) | 13 (7.3%) | |
Graduate or post-graduate | 5 (5.4%) | 4 (2.3%) | |
Professional degree | 1 (1.1%) | 5 (2.8%) | |
Occupation of the head of the family | 0.4 | ||
Skilled worker | 42 (45%) | 69 (39%) | |
Semi-skilled worker | 22 (24%) | 56 (32%) | |
Clerical/ shop owner/ farmer | 9 (9.7%) | 21 (12%) | |
Unskilled worker | 13 (14%) | 14 (7.9%) | |
Professional | 5 (5.4%) | 7 (4.0%) | |
Unemployed | 2 (2.2%) | 6 (3.4%) | |
Semi-professional | 0 (0%) | 4 (2.3%) | |
1Median (Q1, Q3); n (%) | |||
2Wilcoxon rank sum test; Fisher's Exact Test for Count Data with simulated p-value | |||
Income in INR. | |||
Per capita income equally distributes the family income not by gender or age | |||
There are no significant differences between the two types of participants.
The reported education of the head of the family is rarely beyond high school.
Per-capita income and income percentiles
It is difficult to interpret monthly earnings without a reference point. It is more informative to compare the per-capita income of these households with the rest of the population. An established method of doing this (as explained in methodology) is to see which percentile do these patients belong to based on their reported income.
Most of the patients belong to the lowest centiles with respect to per capita income as can be seen from the figure below. For illustration, 65% of the respondents earned less than 90 % of the population.
Comparison of Household Level Variables with NFHS-5 Data
To better situate these patients with respect to the general standard
of living in Bangalore city, we obtained raw data from NFHS-5 via
Department of Health Services, Government of the United States.
We compare key variables related to assets and amenities in our sample
with that of the general population in Bangalore city.
Characteristic | 1. Study Participants | 2. NFHS-5 | p-value2 |
|---|---|---|---|
Household has an air conditioner or cooler? | 1 (0.5%) | 208 (28%) | <0.001 |
Household has a water pump? | 2 (1.1%) | 76 (10%) | <0.001 |
Household has a computer? | 15 (8.2%) | 179 (24%) | <0.001 |
Anyone in the household owns the land of residence or any other residential land? | 91 (50%) | 195 (26%) | <0.001 |
Household has a car? | 8 (4.4%) | 142 (19%) | <0.001 |
Household has electricity? | 180 (99%) | 743 (99%) | 0.7 |
Household has a radio? | 0 (0%) | 32 (4.3%) | 0.004 |
Household has a black and white television? | 1 (0.5%) | 13 (1.7%) | 0.3 |
Household has a colour television? | 175 (96%) | 699 (93%) | 0.10 |
Household has a mattress? | 180 (99%) | 695 (92%) | 0.001 |
Household has a pressure cooker? | 181 (99%) | 721 (96%) | 0.013 |
Household has a chair? | 178 (98%) | 716 (95%) | 0.10 |
Household has a cot or bed? | 176 (97%) | 693 (92%) | 0.024 |
Household has a table? | 167 (92%) | 637 (85%) | 0.012 |
Household has an electric fan? | 171 (94%) | 721 (96%) | 0.3 |
Household has a sewing machine? | 1 (0.5%) | 160 (21%) | <0.001 |
Household has internet access? | 126 (69%) | 503 (67%) | 0.5 |
Household has a washing machine? | 93 (51%) | 390 (52%) | 0.9 |
Household has a refrigerator? | 152 (84%) | 555 (74%) | 0.006 |
Household has a bicycle? | 4 (2.2%) | 220 (29%) | <0.001 |
Household has a motorcycle or scooter? | 126 (69%) | 575 (76%) | 0.051 |
Household has a landline telephone? | 1 (0.5%) | 59 (7.8%) | <0.001 |
Household has a mobile telephone? | 174 (96%) | 726 (96%) | 0.7 |
Household owns agricultural land? | 12 (6.6%) | 51 (6.8%) | >0.9 |
Household owns any livestock? | 3 (1.6%) | 15 (1.9%) | 0.8 |
Anyone in the household has a bank account? | 179 (98%) | 718 (95%) | 0.059 |
1n (%) | |||
2Pearson's X^2: Rao & Scott adjustment | |||
The NFHS data is weighted with distric level weights. Weights for study participants is 1. | |||
Data Acknowledgement: DHS Program USAID | |||
This comparison is interesting. While certain items traditionally mapped by NFHS like radio, internet access, landline may be less relevant in urban settings. Ownership of costly consumer durables like AC, car and computer is definitely lower in our sample. On the other hand posession of land which is a marker of wealth is higher in our sample.
We next compare the two samples on basic amenities and housing.
Characteristic | 1. Study Participants | 2. NFHS-5 | p-value2 |
|---|---|---|---|
What is the total number of people in the household? | 4.00 (3.00, 5.00) | 4.00 (3.00, 4.00) | <0.001 |
Number of sleeping rooms (capped at 4). | <0.001 | ||
1 | 62 (34%) | 10 (1.3%) | |
2 | 99 (54%) | 245 (32%) | |
3 | 21 (12%) | 394 (52%) | |
4+ | 0 (0%) | 105 (14%) | |
Number of persons per sleeping room (capped at 4). | <0.001 | ||
1 | 4 (2.2%) | 299 (40%) | |
2 | 61 (34%) | 392 (52%) | |
3 | 82 (45%) | 49 (6.5%) | |
4+ | 35 (19%) | 14 (1.8%) | |
Religion of the household | <0.001 | ||
christian | 8 (4.4%) | 30 (4.0%) | |
hindu | 47 (26%) | 650 (86%) | |
muslim | 127 (70%) | 67 (8.9%) | |
other | 0 (0%) | 6 (0.9%) | |
Household uses lpg or natural gas for cooking? | 182 (100%) | 728 (97%) | 0.013 |
Type of floor in the house | 0.14 | ||
Finished Material | 182 (100%) | 745 (99%) | |
Unfinished Material | 0 (0%) | 9 (1.2%) | |
Type of wall in the house | <0.001 | ||
Finished Material | 181 (99%) | 644 (85%) | |
Unfinished Material | 1 (0.5%) | 110 (15%) | |
Type of roof in the house | 0.015 | ||
Finished Material | 182 (100%) | 729 (97%) | |
Unfinished Material | 0 (0%) | 25 (3.3%) | |
1Median (Q1, Q3); n (%) | |||
2Design-based KruskalWallis test; Pearson's X^2: Rao & Scott adjustment | |||
The NFHS data is weighted with distric level weights. Weights for study participants is 1. | |||
Data Acknowledgement: DHS Program USAID | |||
It is clear that the houses of study participants are in general smaller than an average household. While, less than 40% of representative households have less than 3 rooms for sleeping, more than 80% of study participants lived in a house with 1 or 2 rooms.
This limited space is shared by more members in study participants’ houses as compared to a representative household.
The distribution of religion in the study group also differs significantly from the general population as surveyed two years back.
On the other hand, the households of study participants are no different or even better in amenities like cooking fuel and type of house construction.
The difference in household size is difficult to interpret due to overlapping medians and first quantile. We therefore, examined the distribution to understand this.
We can see that this difference is mainly due to a high proportion of study participants having 5 or more household members as compared to the representative household based on NFHS 5 data.
Comparison of Individual Level Variables with NFHS-5 Data
Characteristic | 1. Study Participants | 2. NFHS-5 | p-value2 |
|---|---|---|---|
Age | 24 (22, 27) | 33 (26, 38) | <0.001 |
Years of education | 9.0 (7.0, 10.0) | 12.0 (10.0, 15.0) | <0.001 |
Unknown | 1 | 0 | |
1Median (Q1, Q3) | |||
2Design-based KruskalWallis test | |||
The NFHS data is weighted with distric level weights. Weights for study participants is 1. | |||
Only male respondents of age 18-45 years in NFHS-5 dataset are used to ensure comparability of samples | |||
Data Acknowledgement: DHS Program USAID | |||
Patients are significantly younger than the median adult male represented by NFHS-5 data.
On average they are also significantly and substantially (3 years difference) less educated than the median male of the city.
Neighbourhood Level Relative Wealth Index
As described in the methodology, RWI is a new and unitless measure and thus we first examined its relevance for the whole country. As seen in the figure below, at the country level the RWI is high in capital and large cities as compared to other areas. This distribution is in line with our expectation and thus we went on to analyse the RWI for Bangalore district.
At the city level, we see higher RWI values (on a scale of 0-1) in central buisiness district and other hubs of economic activity (Figure below).
However unlike comparison with NFHS median household we see no evidence of locality level deprivation in study participants. In the figure below we see that the residential address of most participants are in high RWI grid tiles.
Visually this finding is so apparent that we did not undertake comparison of median RWI’s of the participant’s localities with the median RWI of other tiles.
Neighbourhood Level Measures of Population Density and Overcrowding
As described in the methodology, we first aggregated the number of persons in square grids of 1 km side. This yields population density in persons per km2 of built-up area. Since, this is a relatively new measure, similar to RWI we first examined the data at the country level.
As can be seen in the figure, at the state level UP, Bihar, West Bengal and Kerala have an overall higher pop. density than most other states. This is in line with the population projections from various sources. Furthermore, major urbanisation centres like Delhi, Bangalore, Mumbai and almost all state capitals show highest population densities.
However, since there is no established index of overcrowding using number of persons per km square of built-up area we have to look at this data in a relative fashion. We have used percentile ranking to better discern areas where pop. density is substantially higher than the rest of the country. For illustration, we have 2228908 (22 lakh) individual grid cells, each being a 1 km2 peice of land with some built up structures. If we arrange them in ascending order of pop.density we can rank them from the least populated to most populated. Finally, we can calculate the percentile rank by dividing the rank with the number of grid cells. This leads to an easily interpretable metric for comparison.
In the figure below we see that extremely high values are seen only in some urban centres. We then examined the data for Bangalore.
It is striking to note that Bengaluru Urban district has one of the most densely populated areas in the country. We see large areas that are in the top 0.01 percent in terms of the number of persons per sq. km of built-up structures.
When we examine the residential address of participants with respect to the pop. density it is visible that many of them come from the most densely populated areas - not only in the city but in the whole country (Figure below).
Nevertheless, it is not clear if the patients experience statistically significant levels of overcrowding than an average person in Bangalore. To test this hypothesis we compare the pop. density at the residential addresses of participants with the overall pop. density distribution in Bangalore using KS test as detailed in methodology.
We find that a higher proportion of study participants lived in areas with higher pop. density than would be expected if they were a random subsample from the population of Bangalore. This difference is significant and the magnitude can be judged from the medians and quantiles given below.
Characteristic | All 1km.sq grid cells in Bangalore | Grid cells with study participants residence | p-value2 |
|---|---|---|---|
Population Density | 537.0 (240.0, 1339.0) | 1487.0 (1231.0, 26093.0) | <0.001 |
1Median (Q1, Q3) | |||
2Wilcoxon rank sum test | |||
Pop.density = persons/km.sq of built-up area | |||
Substance Use Profile
Current and lifetime use of substances
Tables below summarises the lifetime and current use of substances in the sample.
Characteristic | Community | Convenience | p-value2 |
|---|---|---|---|
Tobacco products (cigarettes, bidis, gutkha, zarda etc) | 93 (100%) | 177 (100%) | |
Alcoholic beverage (rum, whisky, beer, tadi, sarai, tharra) | 55 (59%) | 63 (36%) | <0.001 |
Cannabis (ganja, bhang, charas) | 93 (100%) | 144 (81%) | <0.001 |
Opioids (heroin, chitta, tidigesic, fortwin, ultracet, tramazac, tydol, cough syrup) | 93 (100%) | 177 (100%) | |
Sedatives (nitravat, n10, valium, alprax, restyl, zolpid, traika) | 39 (42%) | 56 (32%) | 0.092 |
Amphetamine type stimulants (md, crystal, ice, ecstasy) | 71 (76%) | 90 (51%) | <0.001 |
Cocaine (crack, coke) | 7 (7.5%) | 7 (4.0%) | 0.3 |
Inhalants (correction fluid, whitener, dendrite, puncture fluid, paint thinner, solution, petrol) | 6 (6.5%) | 12 (6.8%) | >0.9 |
Hallucinogens (lsd, acid, stamp, k, ketamine, mushrooms) | 5 (5.4%) | 6 (3.4%) | 0.5 |
1n (%) | |||
2Pearson's Chi-squared test; Fisher's exact test | |||
Characteristic | Community | Convenience | p-value2 |
|---|---|---|---|
Tobacco products (cigarettes, bidis, gutkha, zarda etc) | 92 (99%) | 177 (100%) | 0.3 |
Alcoholic beverage (rum, whisky, beer, tadi, sarai, tharra) | 45 (82%) | 18 (29%) | <0.001 |
Cannabis (ganja, bhang, charas) | 89 (96%) | 70 (49%) | <0.001 |
Opioids (heroin, chitta, tidigesic, fortwin, ultracet, tramazac, tydol, cough syrup) | 93 (100%) | 177 (100%) | |
Sedatives (nitravat, n10, valium, alprax, restyl, zolpid, traika) | 22 (56%) | 11 (20%) | <0.001 |
Amphetamine type stimulants (md, crystal, ice, ecstasy) | 60 (85%) | 29 (32%) | <0.001 |
Cocaine (crack, coke) | 2 (29%) | 1 (14%) | >0.9 |
Inhalants (correction fluid, whitener, dendrite, puncture fluid, paint thinner, solution, petrol) | 1 (17%) | 0 (0%) | 0.3 |
Hallucinogens (lsd, acid, stamp, k, ketamine, mushrooms) | 1 (20%) | 0 (0%) | 0.5 |
1n (%) | |||
2Fisher's exact test; Pearson's Chi-squared test | |||
Characteristic | Community | Convenience | p-value2 |
|---|---|---|---|
Tobacco products (cigarettes, bidis, gutkha, zarda etc) | 93 (100%) | 177 (100%) | |
Alcoholic beverage (rum, whisky, beer, tadi, sarai, tharra) | 55 (59%) | 63 (36%) | <0.001 |
Cannabis (ganja, bhang, charas) | 93 (100%) | 144 (81%) | <0.001 |
Opioids (heroin, chitta, tidigesic, fortwin, ultracet, tramazac, tydol, cough syrup) | 93 (100%) | 177 (100%) | |
Sedatives (nitravat, n10, valium, alprax, restyl, zolpid, traika) | 39 (42%) | 56 (32%) | 0.092 |
Amphetamine type stimulants (md, crystal, ice, ecstasy) | 71 (76%) | 90 (51%) | <0.001 |
Cocaine (crack, coke) | 7 (7.5%) | 7 (4.0%) | 0.3 |
Inhalants (correction fluid, whitener, dendrite, puncture fluid, paint thinner, solution, petrol) | 6 (6.5%) | 12 (6.8%) | >0.9 |
Hallucinogens (lsd, acid, stamp, k, ketamine, mushrooms) | 5 (5.4%) | 6 (3.4%) | 0.5 |
1n (%) | |||
2Pearson's Chi-squared test; Fisher's exact test | |||
ASSIST Substance Use Risk Scores
Characteristic | Community | Convenience | p-value2 |
|---|---|---|---|
Risk category: tobacco | 0.3 | ||
High risk | 0 (0%) | 4 (2.3%) | |
Moderate risk | 93 (100%) | 173 (98%) | |
Risk category: alcohol | <0.001 | ||
High risk | 1 (1.1%) | 3 (1.7%) | |
Low risk | 70 (75%) | 169 (95%) | |
Moderate risk | 22 (24%) | 5 (2.8%) | |
Risk category: cannabis | <0.001 | ||
High risk | 29 (31%) | 10 (5.6%) | |
Low risk | 9 (9.7%) | 109 (62%) | |
Moderate risk | 55 (59%) | 58 (33%) | |
Risk category: cocaine | 0.12 | ||
Low risk | 91 (98%) | 177 (100%) | |
Moderate risk | 2 (2.2%) | 0 (0%) | |
Risk category: opioids | <0.001 | ||
High risk | 68 (73%) | 177 (100%) | |
Moderate risk | 25 (27%) | 0 (0%) | |
Risk category: stimulants | <0.001 | ||
High risk | 3 (3.2%) | 1 (0.6%) | |
Low risk | 49 (53%) | 159 (90%) | |
Moderate risk | 41 (44%) | 17 (9.6%) | |
Risk category: hallucinogens | |||
Low risk | 93 (100%) | 177 (100%) | |
Risk category: sedatives | <0.001 | ||
Low risk | 81 (87%) | 173 (98%) | |
Moderate risk | 12 (13%) | 4 (2.3%) | |
Risk category: inhalants | >0.9 | ||
Low risk | 92 (99%) | 174 (98%) | |
Moderate risk | 1 (1.1%) | 3 (1.7%) | |
1n (%) | |||
2Fisher's exact test; Pearson's Chi-squared test | |||
- Problematic use of all classes of substances except opioids and tobacco is significantly and substantially higher in sample recruited from the community.
This difference is more clearly shown in the figure below where we compare the proportion of problematic use (ASSIST categories: mild risk and high risk) in the two samples.
Age at onset of Problematic Substance Use
When we compare the age at onset of problematic substance use for various classes, we see no significant differences between the two groups (Table below).
Characteristic | N | Community1 | Convenience1 | p-value2 |
|---|---|---|---|---|
Age At Onset: Tobacco | 270 | 16 (14, 18) | 16 (14, 18) | 0.3 |
Age At Onset: Cannabis | 152 | 17 (15, 20) | 17 (15, 18) | 0.3 |
Age At Onset: Opioids | 270 | 21 (19, 25) | 20 (18, 23) | 0.056 |
Age At Onset: Alcohol | 31 | 20 (17, 24) | 16 (15, 19) | 0.051 |
Age At Onset: Stimulants | 62 | 19 (18, 22) | 19 (18, 20) | >0.9 |
Age At Onset: Sedatives | 18 | 20 (18, 22) | 15 (15, 16) | 0.002 |
1Median (Q1, Q3) | ||||
2Wilcoxon rank sum test | ||||
Classes with n < 10 are not shown | ||||