Table 1 Categorisation of explanatory variables

Calendar year of entry

Ordered categorical:

Events, program performances usually vary with time

2005, 2006, 2007, 2008, 2009, 2010

Current age groups (years)

Ordered categorical: 15 – 24, 25 – 34, 35 – 44 , >45 (using Lexis expansion)

Events vary with age and adjusting for age attained in cohort is more appropriate than by age at entry as this avoids residual confounding

Gender

Binary: Females, Males

Gender is usually associated with most diseases and thus a strong confounder

Region or province

Unordered categorical: Littoral, South-West, West, Other

Risk of clustering of events by place of residence

Marital status

Unordered categorical: Single, Monogamous, polygamous, divorced, widowed

Socio-cultural and economic empowerment is usually differential in African context

Partner HIV Status

Unordered categorical: Negative

Health seeking behaviour, adherence, treatment success or failure may be determined by partner HIV status, viral load and viral strain

Positive but not taking ART, Positive and taking ART,

Unknown

Occupation

Binary: Lower grade, higher grade

Proxy measure of level of socio-economic status. Grading based on International Standard Classification of Occupations (ISCO)

Distance(km)

Binary: ≤5, >5

Usual walking distance within 30 minutes to health facility is 5 km (indicator of accessibility); also reflects the population living in the urban centre and usually accessible to community workers

Alcohol Intake

Binary: No, Yes

Interaction with drugs, co-morbidity, behaviour change

Smoking

Binary: No, Yes

Factor of many co-morbid conditions

WHO Clinical Stage

Ordered categorical:

Risk of death and health care seeking behaviour is influenced by the severity of disease usually; the staging also guides when to start ART

I = Asymptomatic condition

II = Mild

III = Advanced

IV = Severe

Immune deficiency

Ordered categorical: based on CD4 count

Risk of opportunistic infection and thus death depend on CD4 count; initiation of ART also depends on CD4 count

None: ≥500

Mild: 350-499

Advanced: 200-350

Severe: <200

Haemoglobin level (g/dl)

Binary: <10, ≥10

Cut off point based on the median value in this HIV cohort to define anaemia

Alanine amino-Transferase, ALAT (IU/l)

Binary: <50 , ≥50

Cut off point as determined by the hospital laboratory

Above which indicates liver injury (more specific marker)

Aspartate amino-Transferase, ASAT (IU/l)

Binary: <45, ≥45

Cut off point as determined the hospital laboratory

Above which indicates liver injury(less specific marker)

Fasting blood sugar (mg/dl)

Binary: <126, ≥126

Cut off point above which defines diabetes mellitus

Creatinine level(mg/l)

Binary: ≤15, >15

Cut off point as determined the hospital laboratory

Above which indicates kidney injury

NNRTI Regimen

Binary: EFV(efavirenz)-based, NVP(nevirapine)-based

EFV or NVP is present in all first line regimens; their relative efficacy, tolerance or toxicity may be relevant

NRTI Regimen

Unordered categorical: ABC(abacavir)/TDF(tenofovir)-based, AZT(zidovudine)-based,

NRTI form the backbone of all first line regimens; while 3TC is invariably present, the relative toxicity, efficacy and tolerance may be relevant to adherence and emergence of drug resistance and thus to Mortality and LTFU

D4T(stavudine)-based, None (missing value)

Drug Change

Binary: No, Yes

Proxy measure for the presence of drug toxicity (usually) or drug resistance (rarely)

Cotrimoxazole Prophylaxis

Binary: No, Yes

Measure to prevent common opportunistic infections