Household food insecurity in urban slums and its association with nutritional status of under-five children in Salem District, Tamil Nadu

Study area

Selected notified slums of Salem Corporation, Salem district, Tamil Nadu

Study population

Principal caregivers (≥18 yr) responsible for food preparation/distribution for ≥2 meals/day and ≥5 days/week and eldest under-five child in the family if present.

Inclusion criteria

Principal caregivers in family households (members related by marriage, birth, or adoption); joint families considered single units; informed consent required.

Exclusion criteria

Non-family households, locked houses, or unavailable respondents after three visits.

Sample size

Based on 16.7% severe food insecurity prevalence in urban areas (Anand et al¹²), 95% CI, 5% precision, design effect 1.5, and 10% non-response, yielding 360 households minimum.

Sampling method

Multistage random sampling. From Tamil Nadu districts with highest slum prevalence (Chennai, Coimbatore, Salem, Trichy, Madurai), Salem was chosen.¹³ Of 246 slums in 60 wards, 5 wards and one slum per ward were selected randomly.¹⁴ Households were proportionately sampled; every 4th household chosen via systematic random sampling, starting with spin-a-pen random walk. If criteria unmet, adjacent household selected. Door-to-door survey covered 363 households.

Study instruments

Interviewer-administered questionnaire on socio-demographics, diet, food insecurity, and family history. Regular public distribution system (PDS) use defined as ≥1 ration collection/month for 3 months; government aid included Integrated Child development scheme (ICDS), mid-day meals, pensions, insurance, or welfare schemes. Food consumption was recorded as the amount of raw cereals consumed by all household members, using premeasured utensils for conversion to grams. Daily intake was first calculated in grams/day and then multiplied by 30 to obtain the monthly equivalent. Community-level variables included health services, markets, safe water, Liquified Petroleum Gas (LPG). Household food insecurity was assessed and categorised using household food insecurity access scale (HFIAS)¹⁵ and household food insecurity access prevalence (HFIAP) tools. Anthropometry was (height, weight/length) recorded for under-five children; eldest child was selected if multiple children were present.

Household food insecurity categorisation and prevalence

Food secure households (Category 1) reported no insecurity (Q1a/Q1=0; Q2–Q9=0). Mildly food insecure households (Category 2) showed anxiety or concern about food (affirmative responses to Q1a–Q4a) but no limitations in food quantity or quality (Q5–Q9=0). Moderately food insecure households (Category 3) reported compromises in food quality or quantity (affirmative responses to Q3a–Q6) without severe reductions in intake (Q7–Q9=0). Severely food insecure households (Category 4) experienced serious disruptions in eating patterns or hunger (affirmative responses to Q5a–Q9a). Prevalence was calculated as the percentage of households in each HFIA category, for example, severe food insecurity was estimated by dividing the number of households in Category 4 by the total assessed and multiplying by 100, as per the HFIAS indicator guide.¹⁵

Nutritional status

Under-five children were assessed by anthropometric measurements like weight and height/length. Nutritional status was assessed using age appropriate WHO growth chart.¹⁶ Classification into normal, moderately wasted, and severely wasted categories was based on WHO weight-for-height Z-score definitions: normal (≥ −2 SD), moderate wasting (between −2 SD and −3 SD), and severe wasting (< −3 SD).¹⁶

Quality control measures

A pilot study in Sivathapuram block PHC (Urban), Salem district, with 35 participants validated the Tamil HFIAS translation and assessed feasibility. Data collection used pretested tools, interviewer-administered questionnaires, and calibrated anthropometric instruments (weight±100 g; height±0.1 cm). Double weighing was used for non-cooperative children, and privacy ensured during interviews and assessments. Pilot results were excluded from main results.

Data management

Data were entered in Microsoft Excel and analysed using SPSS v16 (IBM Corp., Arm, NY, USA). Weight-for-height (WHZ)—were derived using WHO growth charts. Data cleaning excluded incomplete entries, and coding was applied prior to analysis.

Statistical tools

Descriptive statistics [proportions, means±standard deviation (SD)] were calculated. Associations were tested using chi-square and Fisher’s exact tests. One-way ANOVA assessed food group differences across HFIAS categories. Multivariate regression identified predictors of food insecurity with P<0.05 taken as significant.

Prevalence and Predictors of household food insecurity

The study of 363 households revealed marked socio-demographic and health vulnerabilities. All caregivers were female and nearly half aged 18–34 yr and mostly homemakers, while household heads were predominantly males, aged 25–54 yr and employed as unskilled/skilled workers. Most caregivers reported no income and lived in rented homes which were largely semi-pucca. The majority were Hindu, married couples living in joint or nuclear families, typically small in size (≤4 members). Health vulnerabilities were evident, with one-third reporting multiple co-morbidities and nearly half having dependents ( Table I).

Food insecurity was widespread, affecting 82.1% of households (27.5% mild, 34.4% moderate, 20.1% severe); >80% reported anxiety and reduced food quality/diversity, and 18.7% had members go a whole day without food. Education was protective, with graduates significantly more food secure compared to those with middle school education (P<0.001). Homemakers and female-headed households were highly vulnerable, showing greater levels of moderate to severe insecurity (P<0.001). Economic and structural factors were decisive: households with higher income (>₹10,000/month) were more secure, while those with no income were severely insecure (P<0.001). Single-earner families had greater moderate to severe insecurity, whereas multiple-earner households were more secure (P<0.001). Larger families and those simultaneously caring for under-five children and elderly members were especially at risk (P<0.001). The presence of co-morbidities increased insecurity, with affected families showing higher levels of moderate and severe categories (P<0.001). Environmental deprivation further compounded risks. Kutcha housing, reliance on private water supply (P<0.001), lack of refrigerator (P<0.001), firewood use (P<0.001), irregular ration card use, and absence of government aid were all significantly associated with severe food insecurity ( Table I).

All variables significant at P<0.001 were dichotomised for further analysis. Covariates—family size, dependents, house ownership, age and gender of household head, co-morbidities, and public distribution system (PDS) usage—were included in the regression model ( Table II). Multivariate logistic regression explained 39.6% of variability in food insecurity. Significant factors included small family size (≤4 members) which was protective against mild and moderate insecurity, while rented accommodation markedly increased risk across mild, moderate and severe categories. Co-morbidities elevated severe risk, whereas regular PDS use was protective. Younger households (<37 yr) had reduced odds of moderate and severe food insecurity, while male-headed households were less likely to experience severe insecurity. Mean monthly food consumption of various food constituents was significantly reduced among insecure households ( Table III). Cereals, pulses, oil, and vegetables showed consistent declines with increasing insecurity.

Children’s nutritional status and its link to food insecurity

Among 169 under-five children assessed, n=45 (26.6%) were aged 0–12 months, n=35 (20.7%) aged 12–24 months, and n=89 (52.7%) aged 2–5 years; boys comprised 58.6% (n=99) and girls 41.4% (n=70). Anthropometric measures showed a mean height of 82.4±12.5 cm and mean weight of 9.2±2.88 kg and 80 (47.3%) were classified as wasted. Household food insecurity was significantly associated with nutritional status (P< 0.001), with children from insecure households nearly three times more likely to be undernourished [Odds ratio (OR)=2.75; 95% confidence interval (CI): 1.31–5.75]. Normal weight-for-height declined with increasing insecurity: 31 (34.8%) in secure, 40 (44.9%) in mild, 17 (19.1%) in moderate, and 1 (1.1%) in severe households. Severe wasting was concentrated in the most deprived families, affecting 29 (52.7%) compared to 17 (30.9%) in moderate, 3 (5.5%) in mild, and 6 (10.9%) in secure households.