Assessment of the Healthy Lifestyle Profile of Adolescents: A School‐Based Cross‐Sectional Study in Bahawalpur, Pakistan Authors: Rameesha Shauket, Anum Safeer, Rida Ali, Humaira Mahmood, Farrah Pervaiz, Inibehe Ime Okon Published: 2026 ABSTRACT ABSTRACTBackgroundAdolescence is a critical period for the formation of health behaviors, yet composite lifestyle data from rural districts of Pakistan are scarce. We described the healthy lifestyle profile of school‐going adolescents in Bahawalpur, South Punjab, and its association with socio‐demographic factors.MethodsThis descriptive cross‐sectional study enrolled 319 adolescents aged 11–19 years from four public schools in one randomly selected tehsil during August 2024. Two over‐age records were excluded, leaving 317 eligible participants for descriptive tabulation; three of these had a missing item and were further excluded, leaving 314 participants for the complete‐case composite analysis. Lifestyle was measured with a nine‐item behavioral composite informed by the Adolescent Lifestyle Profile. We treated the composite as a formative index with a possible range of 0–38 and analyzed it mainly as a continuous score. We assessed associations with Mann–Whitney U tests (reporting rank‐biserial effect sizes), Spearman correlation, χ 2 tests with Cramér's V, and multivariable linear regression with heteroscedasticity‐consistent (HC3) standard errors.ResultsThe mean composite score was 21.36 (SD 3.79; median 22, IQR 19–24; observed range 11–31). When scores were grouped into pragmatic descriptive bands, most adolescents fell in the middle of the distribution, and none reached the highest band. Girls scored modestly higher than boys (21.83 vs. 20.81; Mann–Whitney p = 0.0075; rank‐biserial r = 0.175). In multivariable regression (R 2 = 0.073), grade was the only independent predictor of a higher score (β = 0.78, 95% CI 0.34–1.22, p < 0.001). Age, sex, parental education, family size, and household income were not.ConclusionAdolescents attending these four schools reported largely mid‐range lifestyle behaviors. Screen time was high, fruit and vegetable intake low, sugary‐drink consumption frequent, and vigorous physical activity limited. School‐based health promotion that targets these specific behaviors and accounts for grade‐ and sex‐related differences is warranted. The findings apply to students of the four sampled schools and should not be generalized to all adolescents in the district.Keywords: adolescents, health behavior, healthy lifestyle, Pakistan, school‐based study Introduction Adolescence, spanning ages 10–19 years, is a critical developmental period during which health‐related behaviors take shape and tend to persist into adulthood [1]. The patterns of diet, physical activity, screen use and sleep formed in these years influence the long‐term risk of non‐communicable diseases, a burden that continues to grow in low‐ and middle‐income countries [2]. An estimated 1.3 billion adolescents live worldwide, and a large share of Pakistan's population is aged 10– 19 years [3]. In low‐ and middle‐income countries, adolescents face pressures that shape their lifestyle choices, including rapid urbanization, dietary transition and socio‐economic constraints [4]. School‐based studies are well placed to describe these behaviors because they reach a concentrated and accessible population [5, 6]. Unhealthy diets, physical inactivity and cardiovascular risk factors among adolescents have been documented in Pakistan, mostly in the larger cities [7, 8]. Evidence from rural districts of South Punjab is limited, and few studies have combined several behaviors into a single composite profile [9, 10, 11]. This gap matters for two reasons. Composite profiling captures the clustering of risk behaviors that single‐behavior studies miss, and rural, lower‐income populations may differ in systematic ways from the urban samples that dominate the literature. Bahawalpur, a district of South Punjab with relatively low household income and educational attainment, is under‐represented in adolescent‐health research. Describing the lifestyle behaviors of its school‐going adolescents, and the factors associated with them, can inform locally tailored, gender‐sensitive health promotion. This study aimed to describe the lifestyle behaviors of school‐going adolescents in one tehsil of Bahawalpur and to examine their association with socio‐demographic factors: sex, grade, parental education, family size, and household income. Methods Study Design and Setting We conducted a descriptive cross‐sectional study in four public schools of Bahawalpur district, South Punjab, Pakistan. The four schools spanned different settings: one urban, one peri‐urban, and two rural. The sample is therefore not exclusively rural. Data were collected in August 2024. The study is reported in line with the STROBE guidance for cross‐sectional studies. Participants and Eligibility Eligible participants were adolescents aged 11–19 years attending the selected schools. Participants who declined to take part or returned incomplete questionnaires on the composite items were not included in the composite analysis. There were no health‐based exclusions. Sample Size The minimum sample size was calculated using n = Z 2 P(1 − P)/e 2, with Z = 1.96 (95% confidence), an expected prevalence of poor self‐perceived health of 0.29 from a comparable Pakistani adolescent study [11], and a margin of error e = 0.05, giving a minimum of 316. A total of 319 adolescents were enrolled. No design effect was applied for the multistage design; this is acknowledged as a limitation. Sampling and Participant Flow A multistage sampling design was used. In the first stage, the tehsils of Bahawalpur district formed the sampling frame, and one tehsil was selected using computer‐generated random numbers. In the second stage, four public schools registered with the District Education Authority (EMIS) in the selected tehsil were selected, again using computer‐generated random numbers. In the third stage, students aged 11–19 years were selected from each school's roll‐call list by systematic sampling with a fixed interval of two (i.e., every second student). The starting position (odd‐ or even‐numbered) was randomly determined at the first school, yielding the odd‐numbered positions on the list; the same starting position was then applied at the remaining three schools rather than being re‐randomized independently at each site. All approached students agreed to participate (response rate 100%), and there were no refusals. Of 319 enrolled adolescents, two records with ages recorded as 20 and 21 years fell outside the 11–19‐year eligibility range and were excluded, leaving 317 eligible participants. A further three (0.9%) had a missing response to a single composite item. Because missing data affected fewer than 1% of eligible participants and were confined to isolated single items rather than whole questionnaires, we judged the loss negligible and used complete‐case analysis, which is unlikely to introduce material bias at this level of missingness. We did not formally model the missing‐data mechanism, given the very small number of affected records. The complete‐case sample was therefore 314 for the score, group comparisons, and regression. The number of students on the four roll‐call lists (the eligible denominator) could not be retrieved from school records and is reported as a limitation. Data‐Collection Instrument Data were collected with a structured, self‐administered questionnaire in two sections: (i) socio‐demographics (age, sex, grade, family size, education and occupation of the household head and mother, and monthly household income) and (ii) lifestyle behaviors. The lifestyle items were a study‐specific set informed by the Adolescent Lifestyle Profile (ALP) [12] and related adolescent health‐promotion instruments [13]; they do not constitute a validated ALP score. Eleven items covered fruit and vegetable intake, sugary‐drink consumption, screen time, physical activity, weekly family meals, breakfast frequency, bedtime, wake‐up time, and two self‐reported health‐concern items. Monthly income was classified using the Modified Kuppuswamy Scale (2023) [14], whose income bands are denominated in Indian rupees (INR); these were converted to Pakistani rupees (PKR) using the average interbank USD/INR and USD/PKR exchange rates for July 2024, the month in which the income bands were prepared ahead of data collection (USD/INR ≈ 83.59, USD/PKR ≈ 278.45), giving a cross‐rate of approximately 1 INR = 3.33 PKR. Use of an India‐derived scale and currency conversion are noted as limitations. Healthy Lifestyle Composite Score Each behavioral item was assigned ordinal points so that healthier responses scored higher; screen time, sugary‐drink consumption and bedtime were reverse‐coded accordingly. Nine behavioral items were summed to form a composite score with a possible range of 0–38 (maximum item points are fruit 3, vegetables 3, screen time 4, physical activity 4, sugary drinks 4, family meals 5, breakfast 5, bedtime 5, wake‐up 5). The differences in item maxima were not intended as an importance weighting. They follow directly from the number of ordinal response options offered for each item in the source questionnaire, so an item with six response levels can contribute up to 5 points and an item with four levels up to 3, regardless of the behavior's health importance. We did not apply a theory‐driven weighting scheme, and we make no claim that sleep timing matters more than diet or physical activity. Because all inferential analyses use the continuous score and every component behavior is also reported separately (Tables 3 and 4), the conclusions do not depend on this scaling. The two health‐concern items (0–10) were kept as descriptive variables and were not included in the composite. The full scoring key is provided in Supporting Information S1: Table S1. Table 3Dietary and eating behaviors (n = 317).Behavior n (%)Fruit servings/day1 serving142 (44.8)None103 (32.5)2 servings72 (22.7)Vegetable servings/day1 serving194 (61.2)2 servings97 (30.6)None26 (8.2)Sugary drinks/day2 per day124 (39.1)None119 (37.5)3 per day55 (17.4)5+ per day9 (2.8)4 per day9 (2.8)Family meals/week343 (13.6)420 (6.3)513 (4.1)69 (2.8)7134 (42.3)2 or less97 (30.6)Breakfast days/week313 (4.1)410 (3.2)56 (1.9)613 (4.1)7256 (80.8)2 or less19 (6) Dietary and eating behaviors (n = 317). Table 4Physical activity, screen time and sleep behaviors (n = 317).Behavior n (%)Screen time/day5+ h140 (44.2)4 h128 (40.4)3 h37 (11.7)2 h10 (3.2)<=1 h2 (0.6)Physical activity/day30 min107 (33.8)15 min89 (28.1)None45 (14.2)45 min40 (12.6)60+ min36 (11.4)Bedtime10 p.m.115 (36.4)11 p.m.84 (26.6)9 p.m.57 (18)Midnight or later28 (8.9)8 p.m.25 (7.9)<=7 p.m.7 (2.2)Wake‐up time5 a.m.234 (73.8)6 a.m.75 (23.7)7 a.m.6 (1.9)9 a.m.1 (0.3)8 a.m.1 (0.3) Physical activity, screen time and sleep behaviors (n = 317). The primary outcome was the continuous composite score. For descriptive presentation only, scores were grouped into five pragmatic bands (0–15, 16–20, 21–26, 27–31, 32–38). These bands are labeling conventions carried over from the source scoring sheet; they are not derived from a validation study or external guideline, and no substantive interpretation (e.g., that a lifestyle is objectively “excellent”) is attached to them. All inferential analyses used the continuous score. Because the composite sums conceptually distinct behaviors that are not expected to be mutually correlated, it is a formative (index) rather than a reflective (scale) measure. For formative indices, internal‐consistency statistics such as Cronbach's alpha are not meaningful, because the items are defining causes of the construct rather than interchangeable effects of it; validity for such indices rests on content coverage of the relevant behaviors rather than on inter‐item correlation [15, 16, 17]. We therefore report the distribution of each component behavior rather than a reliability coefficient. Statistical Analysis Analyses were performed in SPSS v27, with regression diagnostics confirmed in Python (statsmodels). Categorical data are summarized as counts and percentages and continuous data as means (SD) or medians (IQR). Normality of the composite score was assessed with the Shapiro–Wilk test; as the score departed from normality, between‐group comparisons used the Mann–Whitney U test (with the independent‐samples t‐test for confirmation), and the rank‐biserial correlation was reported as the effect size. Cohen's d is reported for mean differences and Cramér's V for χ 2 tests. The association between grade and the score was examined with Spearman's correlation. Multivariable linear regression modeled the composite score on age, sex, grade, household‐head education, maternal education, family size and household income. Predictor coding was: age in years (continuous); sex (0 = male, 1 = female); grade as the exact grade 9–12 (continuous); and household‐head education (1–7), maternal education (1–6), family size (1–4) and household income (1–7) as ordinal variables. Regression assumptions were examined through variance inflation factors (multicollinearity), the Ramsey RESET test (functional form), the Shapiro–Wilk test on residuals (normality), the Breusch–Pagan test (homoscedasticity), and Cook's distance (influence). Because residuals were non‐normal and heteroscedastic, heteroscedasticity‐consistent (HC3) standard errors were used; to address the reviewer's query about the opposite signs of age and grade, models containing age and grade separately and their correlation were also examined. The bivariate behavior‐by‐subgroup analyses (Table 6) are exploratory. Because 18 cross‐tabulations were examined, isolated p values close to 0.05 are not treated as confirmatory; emphasis is placed on the pattern and magnitude (Cramér's V) of associations rather than on individual significance thresholds. Categories were merged to reduce sparse cells, and where any expected cell remained below five, Monte‐Carlo‐estimated exact p values (20,000 replicates) were computed to confirm the χ 2 conclusions. School‐level clustering could not be adjusted for with only four schools; its likely effect is discussed as a limitation. Two‐sided p < 0.05 was considered statistically significant. The composite‐scoring syntax and analysis code are available for editorial verification on request. Table 6Exploratory associations of individual behaviors with sex and grade (n = 317).BehaviorCategoryMale n (%)Female n (%)Sex: V; p Secondary n (%)Higher‐sec n (%)Grade: V; p Fruit intakeNone44 (30.1)59 (34.5)0.134; p = 0.05954 (25.7)49 (45.8)0.232; p = < 0.0011 serving60 (41.1)82 (48)97 (46.2)45 (42.1)2 servings42 (28.8)30 (17.5)59 (28.1)13 (12.1)Vegetable intakeNone16 (11)10 (5.8)0.353; p = < 0.00121 (10)5 (4.7)0.200; p = 0.0021 serving111 (76)83 (48.5)138 (65.7)56 (52.3)2 servings19 (13)78 (45.6)51 (24.3)46 (43)Screen time5+ h76 (52.1)64 (37.4)0.174; p = 0.02295 (45.2)45 (42.1)0.090; p = 0.464a 4 h52 (35.6)76 (44.4)80 (38.1)48 (44.9)3 h16 (11)21 (12.3)25 (11.9)12 (11.2)<=2 h2 (1.4)10 (5.8)10 (4.8)2 (1.9)Physical activity0–15 min59 (40.4)75 (43.9)0.220; p = < 0.00195 (45.2)39 (36.4)0.201; p = 0.00230 min38 (26)69 (40.4)57 (27.1)50 (46.7)45+ min49 (33.6)27 (15.8)58 (27.6)18 (16.8)Sugary drinksNone35 (24.1)84 (49.1)0.291; p = < 0.00162 (29.7)57 (53.3)0.267; p = < 0.0012/day62 (42.8)62 (36.3)85 (40.7)39 (36.4)3/day34 (23.4)21 (12.3)46 (22)9 (8.4)4+/day14 (9.7)4 (2.3)16 (7.7)2 (1.9)Family meals<=2/week35 (24)62 (36.5)0.370; p = < 0.00169 (33)28 (26.2)0.158; p = 0.0193–5/week60 (41.1)16 (9.4)57 (27.3)19 (17.8)6–7/week51 (34.9)92 (54.1)83 (39.7)60 (56.1)Breakfast<=2/week7 (4.8)12 (7)0.058; p = 0.59214 (6.7)5 (4.7)0.099; p = 0.2093–5/week12 (8.2)17 (9.9)23 (11)6 (5.6)6–7/week127 (87)142 (83)173 (82.4)96 (89.7)Bedtime<=9 p.m.47 (32.2)42 (24.7)0.094; p = 0.42461 (29.2)28 (26.2)0.190; p = 0.01010 p.m.53 (36.3)62 (36.5)72 (34.4)43 (40.2)11 p.m.34 (23.3)50 (29.4)50 (23.9)34 (31.8)Midnight+12 (8.2)16 (9.4)26 (12.4)2 (1.9)Wake‐up time5 a.m.92 (63)142 (83)0.229; p = < 0.001a 160 (76.2)74 (69.2)0.160; p = 0.018a 6 a.m.48 (32.9)27 (15.8)42 (20)33 (30.8)7 a.m.+6 (4.1)2 (1.2)8 (3.8)0 (0) Note: Percentages are column percentages within each sub‐group. Categories were merged to limit sparse cells. V, Cramér's V.Abbreviation: V, Cramér's V. a For crossings with an expected cell < 5, the reported p value was confirmed by a Monte‐Carlo exact test (20,000 replicates); conclusions were unchanged. These analyses are exploratory and were not corrected for multiple comparisons. Exploratory associations of individual behaviors with sex and grade (n = 317). Note: Percentages are column percentages within each sub‐group. Categories were merged to limit sparse cells. V, Cramér's V. Abbreviation: V, Cramér's V. For crossings with an expected cell < 5, the reported p value was confirmed by a Monte‐Carlo exact test (20,000 replicates); conclusions were unchanged. These analyses are exploratory and were not corrected for multiple comparisons. Ethical Considerations The study was approved by the Institutional Review/Ethical Review Committee of the Armed Forces Post‐Graduate Medical Institute/National University of Medical Sciences (Ref: 490‐AAA‐ERC‐AFPGMI, 4 July 2024). The change of study location from Rawalpindi to Bahawalpur was communicated to the committee before data collection and received verbal (oral) approval; a written amendment was not issued, which we report transparently. For participants younger than 18 years, class teachers acting in loco parentis provided written consent, and the students provided written assent; participants aged 18–19 years provided their own written informed consent. Questionnaires were self‐administered anonymously in the classroom; no personal identifiers were recorded on the analytic dataset, and completed forms were stored securely and accessible only to the research team. The study was conducted in accordance with the Declaration of Helsinki. Results Socio‐Demographic Characteristics Of 317 eligible participants, 214 (67.5%) were aged 14–16 years and 171 (53.9%) were female; 210 (66.2%) were in secondary school (grades 9–10) and 107 (33.8%) in higher secondary (grades 11–12). Educational attainment among families was limited: 81 household heads (25.6%) and 158 mothers (49.8%) were illiterate. Most adolescents lived in families of 4–6 members (168; 53%), and household incomes clustered in the lower ranges (Tables 1, 2). Table 1Socio‐demographic characteristics of participants (n = 317).Characteristic n (%)Age, years, mean (SD)15.73 (1.39)11–13 years16 (5)14–16 years214 (67.5)17–19 years87 (27.4)SexFemale171 (53.9)Male146 (46.1)GradeSecondary (9–10)210 (66.2)Higher secondary (11–12)107 (33.8)Family size1–3 members9 (2.8)4–6 members168 (53)7–10 members120 (37.9)More than 10 members20 (6.3) Socio‐demographic characteristics of participants (n = 317). Table 2Household socio‐economic characteristics (n = 317).Characteristic n (%)Education of household headIlliterate81 (25.6)Primary school57 (18)Middle school54 (17)High school81 (25.6)Intermediate or diploma28 (8.8)Graduate11 (3.5)Profession or honors5 (1.6)Maternal educationIlliterate158 (49.8)Primary school71 (22.4)Middle school51 (16.1)High school24 (7.6)Intermediate or diploma8 (2.5)Graduate5 (1.6)Monthly household income (PKR)41,45059 (18.6)41,460–124,33053 (16.7)124,340–207,43061 (19.2)207,440–311,06041 (12.9)311,070–414,68038 (12)414,690–829,58028 (8.8)829,58037 (11.7) Note: PKR, Pakistani rupees. Income bands derived from the Modified Kuppuswamy Scale (2023) [14], originally denominated in Indian rupees (INR), converted to PKR using the average July 2024 interbank cross‐rate of 1 INR ≈ 3.33 PKR (USD/INR ≈ 83.59; USD/PKR ≈ 278.45), the month in which the income bands were prepared ahead of data collection in August 2024. Household socio‐economic characteristics (n = 317). Note: PKR, Pakistani rupees. Income bands derived from the Modified Kuppuswamy Scale (2023) [14], originally denominated in Indian rupees (INR), converted to PKR using the average July 2024 interbank cross‐rate of 1 INR ≈ 3.33 PKR (USD/INR ≈ 83.59; USD/PKR ≈ 278.45), the month in which the income bands were prepared ahead of data collection in August 2024. Dietary and Eating Behaviors Dietary intake was generally inadequate. No daily fruit was reported by 103 adolescents (32.5%), and 194 (61.2%) reported only one serving of vegetables per day. Sugary drinks were common: 124 (39.1%) reported two servings daily, and a further 73 (23.0%) reported three or more. Family‐meal patterns varied: 134 (42.3%) ate together seven times weekly, whereas 97 (30.6%) ate two or fewer family meals (Table 3). Physical Activity and Screen Time Physical activity was limited: 107 adolescents (33.8%) reported about 30 min of daily activity and 45 (14.2%) reported none, while only 36 (11.4%) met 60 min or more. Screen time was high: 140 (44.2%) spent five or more hours per day on screens and 128 (40.4%) spent about 4 h (Table 4). Healthy Lifestyle Composite Score Among the 314 participants with complete data, the composite score departed from normality (Shapiro–Wilk p = 0.0035), with a mean of 21.36 (SD 3.79), median 22 (IQR 19–24), and observed range 11–31 against a possible range of 0–38. Using the descriptive bands, the distribution was: 0–15, 25 (8%); 16–20, 95 (30.3%); 21–26, 174 (55.4%); 27–31, 20 (6.4%); and 32–38, 0 (0%). No participant scored above 31 (Figure 1); because the highest band was empty, this reflects only that no one reached that author‐defined threshold, not an external judgment of lifestyle quality. Distribution of the healthy lifestyle composite score (n = 314). Shaded regions mark the pragmatic descriptive bands (labeling convention only, not validated thresholds). The dashed line marks the mean. The possible score range is 0–38; no participant scored above 31. Girls recorded higher composite scores than boys (girls mean 21.83, SD 4.16, n = 169; boys mean 20.81, SD 3.22, n = 145; Mann–Whitney p = 0.0075; rank‐biserial r = 0.175; mean difference 1.01, 95% CI 0.18–1.85; Cohen's d = 0.27, a small effect). Scores were higher among higher‐secondary than secondary students (22.32 vs. 20.86; Mann–Whitney p = 0.0029; rank‐biserial r = 0.204), and grade correlated weakly but significantly with the score (Spearman ρ = 0.159, p = 0.0047). Factors Associated With the Composite Score In multivariable linear regression (n = 314; R 2 = 0.073; adjusted R 2 = 0.052; Table 5), grade was the only independent predictor of the composite score (β = 0.782, 95% CI 0.34–1.22, p < 0.001). Age, sex, household‐head education, maternal education, family size, and household income were not significantly associated after adjustment. Table 5Multivariable linear regression of socio‐demographic factors on the composite score (n = 314). Predictor (coding) β 95% CI p Age (years)−0.343−0.73 to 0.040.084Sex (female vs. male)0.395−0.44 to 1.230.352Grade (9–12, continuous)0.7820.34 to 1.22< 0.001Household‐head education (ordinal, 1–7)0.129−0.15 to 0.410.365Maternal education (ordinal, 1–6)−0.279−0.68 to 0.120.172Family size (ordinal, 1–4)−0.057−0.72 to 0.600.866Household income (ordinal, 1–7)0.222−0.01 to 0.450.057 Note: Model R 2 = 0.073, adjusted R 2 = 0.052. Heteroscedasticity‐consistent (HC3) standard errors. Coding: age in years; sex 0 = male, 1 = female; grade = exact grade (9–12); household‐head education 1–7; maternal education 1–6; family size 1–4; household income 1–7 (all ordinal, higher = more). Multivariable linear regression of socio‐demographic factors on the composite score (n = 314). Note: Model R 2 = 0.073, adjusted R 2 = 0.052. Heteroscedasticity‐consistent (HC3) standard errors. Coding: age in years; sex 0 = male, 1 = female; grade = exact grade (9–12); household‐head education 1–7; maternal education 1–6; family size 1–4; household income 1–7 (all ordinal, higher = more). The apparent opposite signs of age and grade noted by the reviewer reflect their shared variance rather than an independent age effect: age and grade were moderately correlated (r = 0.591), and when age was modeled alone, it showed essentially no association with the score (β = 0.006, p = 0.968). Variance inflation factors were all low (≤ 1.82), indicating that multicollinearity was not severe. The RESET test was non‐significant (p = 0.646), supporting the linear specification; residuals were non‐normal (Shapiro–Wilk p = 0.0068) and heteroscedastic (Breusch–Pagan p < 0.001), so HC3 robust standard errors were used; and no observation was influential (maximum Cook's distance = 0.025). Bivariate Associations Between Individual Behaviors and Sub‐Groups Exploratory χ 2 analyses (Table 6) showed several small‐to‐moderate associations. By sex, significant differences were seen for vegetable intake (Cramér's V = 0.353, p < 0.001), family meals (V = 0.37, p < 0.001), sugary‐drink intake (V = 0.291, p < 0.001), physical activity (V = 0.22, p < 0.001), wake‐up time (V = 0.229, p < 0.001) and screen time (V = 0.174, p = 0.022); fruit intake did not differ significantly by sex after category merging (p = 0.059). By grade, significant differences were seen for fruit intake, vegetable intake, sugary drinks, physical activity, family meals, bedtime and wake‐up time (all p < 0.05), whereas screen time and breakfast did not differ. Inspection of the cross‐tabulated percentages (Table 6) showed that girls reported healthier vegetable intake, screen time and family‐meal patterns, whereas boys reported more physical activity; directional statements are based on these percentages rather than on Cramér's V alone. Discussion Among the adolescents attending these four public schools in one tehsil of Bahawalpur, lifestyle behaviors were largely mid‐range. Most participants clustered around the middle of the composite distribution and none reached the highest descriptive band. Dietary intake, physical activity and screen habits were often sub‐optimal. Girls scored modestly higher than boys, and grade was the only factor independently associated with the composite score after adjustment. Examining the individual behaviors by sex showed that girls reported healthier vegetable intake (two servings/day in 45.6% of girls vs. 13% of boys), lower screen time (five or more hours/day in 37.4% vs. 52.1%) and more frequent family meals (6–7 times/week in 54.1% vs. 34.9%), whereas boys reported more physical activity (45 min or more/day in 33.6% vs. 15.8%). These sex patterns are consistent with school‐based studies among Pakistani adolescents that attribute them to cultural norms and differing recreational opportunities [18, 19, 20]. Grade was associated with fruit intake, physical activity and other behaviors; because health literacy was not measured, we do not attribute the grade association to it, and note that age composition, curriculum, school differences, selection and residual confounding are alternative explanations. Parental education, family size and household income were not significantly associated with the composite score after adjustment. This absence of association should not be read as evidence that these factors are unimportant: the relatively homogeneous socio‐economic profile of the sample, limited variation in the socio‐economic indicators, measurement error, modest power and unadjusted school clustering could each obscure a true association. We therefore state only that no statistically significant association with the socio‐economic indicators used here was observed after adjustment. Studies in comparable Pakistani settings have likewise reported attenuated or non‐significant socio‐economic associations [21, 22]. We deliberately do not interpret the associations of physical activity and screen time with the composite score as independent findings, because those behaviors are components of the score and such associations are partly built in by construction. Similarly, the two health‐concern items were not summed into the composite, and their exploratory correlations with the score were weak and non‐significant (Spearman ρ = 0.091, p = 0.109; and ρ = 0.095, p = 0.094); we therefore make no claim linking health concern to healthier behavior. The association between grade and score was weak (ρ = 0.159), so grade is not a strong predictor; the modeled age–grade relationship should be read with the collinearity noted above in mind. Once other variables were controlled, no association was detected with parental education, family size or household income in this relatively homogeneous setting. This study has several strengths, including a defined sampling frame, multistage random sampling, an adequate sample size and a transparent, fully reproducible scoring procedure. Limitations should also be acknowledged. First, the cross‐sectional design precludes causal inference. Second, behaviors were self‐reported and thus subject to social‐desirability and recall bias. Third, sampling within a single tehsil and four schools limits generalizability to the wider district and country. Fourth, within the systematic sampling step, the random starting position (odd‐ or even‐numbered) was determined once, at the first school, and then applied to the remaining three schools rather than being re‐randomized independently at each site; this could introduce a shared, correlated selection pattern across schools that an independent random start at each site would have avoided. Fifth, no design effect was applied in the sample‐size calculation, and school‐level clustering could not be adjusted for with only four schools. Sixth, the composite is a formative index of heterogeneous behaviors, so internal‐consistency statistics are not applicable. Seventh, the income classification was adapted from an India‐derived socio‐economic scale. Finally, the descriptive score bands are pragmatic labeling conventions rather than validated categories, and are not used here to support substantive claims. Taken together, the findings support school‐based, multi‐component health‐promotion programmes that target diet, physical activity, screen use and sleep, with attention to the specific behaviors that differed by sex and grade. Because the recommendations rest on the individual behavior findings rather than on the small composite differences, they are best framed around concrete behaviors. Useful targets include increasing fruit and vegetable intake, reducing sugary drinks and screen time, and promoting daily physical activity, rather than around sex or the empty top band as such [23, 24, 25, 26, 27, 28, 29, 30]. Conclusion Adolescents attending four public schools in one tehsil of Bahawalpur reported largely mid‐range lifestyle behaviors. Screen time was high, fruit and vegetable intake low, sugary‐drink consumption frequent, and vigorous physical activity uncommon. Grade was the only socio‐demographic factor independently associated with the composite score; the socio‐economic indicators examined were not. These findings are specific to the sampled schools. They support culturally appropriate, behavior‐focused school health promotion, but should not be generalized to all adolescents in Bahawalpur or to rural adolescents across South Punjab. Author Contributions Rameesha Shauket: conceptualization, data curation, project administration, visualization, validation, methodology, writing – review and editing, writing – original draft, investigation. Anum Safeer: validation, writing – original draft, data curation. Rida Ali: data curation, writing – review and editing, writing – original draft, methodology. Humaira Mahmood: supervision, writing – review and editing. Farrah Pervaiz: supervision, writing – review and editing. Inibehe Ime Okon: supervision, writing – review and editing, investigation. Funding The authors have nothing to report. Ethics Statement The study was approved by the Institutional Review/Ethical Review Committee of the Armed Forces Post‐Graduate Medical Institute/National University of Medical Sciences (Ref: 490‐AAA‐ERC‐AFPGMI, July 4, 2024). The study was conducted in accordance with the Declaration of Helsinki. Consent For participants under 18 years, written consent was provided by class teachers acting in loco parentis together with written student assent; participants aged 18–19 years provided their own written informed consent. Conflicts of Interest The authors declare no conflicts of interest. Transparency Statement The corresponding author affirms that this manuscript is an honest, accurate, and transparent account of the study; that no important aspects have been omitted; and that any discrepancies from the study as planned have been explained. Supporting information Supplementary File Acknowledgments The authors have nothing to report. Data Availability Statement The data and the analysis/scoring syntax that support the findings are available from the corresponding author on reasonable request and can be provided for editorial verification. References Associated Data Supplementary Materials Supplementary File Data Availability Statement The data and the analysis/scoring syntax that support the findings are available from the corresponding author on reasonable request and can be provided for editorial verification.