Diabetes mellitus, type 2; Glycated hemoglobin; Pakistan; Primary health care; Self care; Self-management; Sociodemographic factors

Diabetes mellitus has become one of the defining chronic disease challenges of the present era. Global estimates indicate that approximately 537 million adults were living with diabetes in 2021, a figure projected to reach 783 million by 2045, with the sharpest increases concentrated in low- and middle-income countries [1]. Pakistan now has one of the heaviest burdens in the world, ranking among the top three countries in terms of the absolute number of affected adults, and type 2 diabetes mellitus (T2DM) accounts for the overwhelming majority of these cases [1,2]. The World Health Organization (WHO) continues to identify diabetes as a leading driver of blindness, kidney failure, myocardial infarction, stroke and lower-limb amputation, most of which are preventable when the disease is well controlled [2].

Unlike acute illness, T2DM is managed largely by the patient rather than the clinician. The everyday behaviors that determine outcomes—taking medicines as prescribed, following an appropriate diet, being physically active, monitoring blood glucose, examining the feet, attending laboratory tests and coping with psychological strain—unfold at home and between clinic visits and are strongly shaped by the social and economic conditions in which patients live [3,4,5,6]. These behaviors, collectively termed diabetes self-management, are the practical expression of the recommendations set out in contemporary standards of care and in national standards for diabetes self-management education and support [5,6]. Their importance is not merely theoretical: structured self-monitoring of blood glucose, regular physical activity and systematic foot care each has a demonstrable effect on glycemic control and on the risk of complications [7,8,9].

Where patients obtain their medicines and advice also matters. In much of South Asia, care is fragmented across physicians in outpatient clinics and pharmacists in community and hospital-linked settings, and a growing body of trial evidence shows that interventions delivered by either professional group can improve glycemic control, blood pressure and medication adherence [10,11,12,13,14]. An earlier exploratory study from Lahore compared self-management among patients who were attending physicians with those who were attending pharmacists and reported meaningful gaps in several practices, most notably laboratory monitoring [15]. That work supplied the questionnaire and conceptual starting point for the present study, but it did not model patient-level factors that might explain why some patients manage their condition well and others do not.

Internationally, self-care behavior varies with sociodemographic circumstances. Age, sex, marital status, educational attainment, household income, family structure and disease duration have been linked to self-management in studies from Jordan, Ghana, India, Portugal and elsewhere, although the direction and strength of these associations vary considerably across settings [16,17,18,19]. Psychological factors, particularly diabetes-related distress, further shape whether patients sustain healthy routines [20,21]. In Pakistan specifically, qualitative work suggests that barriers differ by educational level and that structured, patient-centered programs can improve self-efficacy and behavior [22,23]. What is largely missing is a quantitative, determinant-focused analysis of self-management among patients as they actually present for care.

This study was therefore designed with two aims. First, we aimed to describe the proportion of patients reporting each of nine recommended self-management practices among adults with T2DM who visit physicians at outpatient clinics and hospital-linked pharmacies in Sargodha, Punjab, Pakistan. Second, we sought to identify the sociodemographic and clinical characteristics independently associated with adequate self-management. Given the descriptive and hypothesis-generating character of the work, we framed it as an exploratory study and report it in accordance with STROBE recommendations for observational research [24].

2.1. Study design and setting

We conducted an exploratory study over an eight-week period from July to August 2025. Data were collected at five sites in Sargodha, Punjab, Pakistan: outpatient clinics and hospital-linked pharmacies at which people with T2DM commonly present to purchase antidiabetic medicines, to undergo diabetes-related laboratory testing and to consult a physician about their condition. Sites were selected purposively to include settings that had a prescribing physician onsite and that together served patients drawn from a range of socioeconomic backgrounds, so that the sample would reflect the mixed clientele typical of urban and peri-urban diabetes care in the region.

2.2. Sample size

Because the number of people with T2DM attending these sites over the study window could not be determined (no register of patients attending the five sites was available) and was therefore treated as unknown and very large, the sample size was estimated using the Cochran formula for an unknown or very large population, n₀ = Z²p(1−p)/e² [25,26]. With a 95% confidence level (Z = 1.96), an expected proportion of 0.50 and a 5% absolute margin of error (e = 0.05), the minimum required sample size was 384. The most conservative value, p = 0.50, was used because no reliable local estimate of overall adherence was available and the nine practices were expected to vary widely in frequency. To allow for up to 20% incomplete or unusable questionnaires, the recruitment target was inflated to 480 (384/0.8), and 483 complete interviews were obtained. This calculation served the primary, descriptive objective of estimating the proportion of patients reporting each practice. Its adequacy for the secondary, analytical objective was checked separately: with 141 patients in the smaller outcome category and 10 parameters in the multivariable model, there were approximately 14 outcome events per parameter, above the commonly used minimum of 10, although this rule of thumb is only a rough guide [27].

2.3. Sampling and recruitment

Participants were enrolled by consecutive sampling. During prearranged data-collection sessions at each site, every patient presenting for diabetes-related medicines, testing or consultation was screened for eligibility, and all those who met the criteria and consented were interviewed in the order in which they presented until the target across sites was reached. This walk-in, consecutive approach is well suited to the exploratory objective and to the flow of patients through pharmacy and clinic settings; its nonprobability nature is acknowledged among the study limitations.

2.4. Eligibility criteria

Adults aged 18 years or older with a physician-confirmed diagnosis of T2DM of at least two years' duration who were attending one of the five study sites in person during the study period and who were willing to provide written informed consent were eligible. Patients with type 1 or gestational diabetes, those newly diagnosed (less than two years), pregnant women and individuals with a psychiatric or cognitive condition that precluded a reliable interview were excluded. These criteria are consistent with those of the source study and with the observed data, in which all participants were adults and reported a disease duration of two years or more [15].

2.5. Questionnaire development

The instrument was adapted from the previously published regional questionnaire used by Malik and colleagues, which was informed by established self-care measures [15]. It comprises two sections. The first recorded sociodemographic and clinical characteristics: sex, age, marital status, years of formal education, monthly household income, family system, duration of T2DM and family history of diabetes. The second captured self-management using ten dichotomous (yes/no) items: cigarette smoking (a risk behavior) and nine recommended practices, namely daily feet checking, daily medication use, exercising for at least 20–30 minutes on at least five days per week, eating a well-balanced planned diet, checking random blood glucose at least once every three months, checking home blood sugars according to physician advice, checking glycated hemoglobin (HbA1c) every three months, using stress-management techniques, and regular random blood pressure monitoring. The content was reviewed by a physician and a pharmacist for clarity and local relevance, the instrument was translated into Urdu and back-translated to confirm fidelity, and it was pretested on a small group of patients who were not included in the final sample. Minor wording revisions followed the pilot test.

2.6. Data collection

Trained interviewers administered the questionnaire face-to-face in a private area at each site. Each interview lasted approximately ten minutes and was conducted separately from any clinical or dispensing encounter so that the responses were not influenced by the consultation itself. Interviewers recorded responses directly, and completed forms were checked for completeness on the same day, which is reflected in the absence of missing values in the final dataset.

2.7. Variables and outcome definition

The nine recommended practices were analyzed both individually and as a summed practice-count index ranging from 0 to 9. Because diabetes self-care behaviors are known to be only weakly intercorrelated, we first examined the internal structure of the nine items. Internal consistency was low (Kuder–Richardson 20 = −0.11), the mean absolute interitem phi coefficient was 0.08, and the Kaiser–Meyer–Olkin measure of sampling adequacy was 0.56, indicating that the items did not form a unidimensional scale. We therefore treated the practices primarily as distinct behaviors and retained the count only as a descriptive summary. For determinant modeling, adequate self-management was defined a priori as adherence to at least five of the nine practices (that is, a majority of recommended behaviors), in line with the common convention of classifying self-care as good when at least half of the maximum possible score is attained [28]; this cutoff also coincided with the median practice count in the sample. Patients below this threshold were classified as having inadequate self-management. Cigarette smoking was treated as an explanatory clinical/behavioral characteristic rather than as a self-care practice.

2.8. Statistical analysis

Continuous variables were summarized as mean and standard deviation or median and interquartile range, and categorical variables were summarized as frequencies and percentages. The proportion of patients reporting each practice was reported with 95% Wilson confidence intervals (CIs). Bivariate associations between patient characteristics and adequate self-management were tested using the chi-square test (with Cramér's V as an effect size) for categorical variables and the independent-samples t test (with Cohen's d) for continuous variables; the Mann–Whitney U test was used as a robustness check for skewed continuous variables. All characteristics were then entered simultaneously into a multivariable binary logistic regression model with adequate self-management as the outcome, and adjusted odds ratios (aORs) with 95% CIs were derived. Model performance was assessed using the likelihood ratio test, the Nagelkerke pseudo-R², the Hosmer–Lemeshow goodness-of-fit test and the area under the receiver operating characteristic curve (AUC). As a sensitivity analysis, a Poisson regression model was fitted to the practice-count index. Given the exploratory design, no adjustments were made for multiple comparisons, and associations were interpreted as hypothesis-generating. A two-sided p value ≤ 0.05 was considered statistically significant. Analyses were performed in Python 3 using the statsmodels (version 0.14) and SciPy (version 1.17) libraries.

2.9. Ethical considerations

The study protocol was reviewed and approved by the Ethical Review Committee, Rai Foundation Medical College, A Project of Rai Institute Medical Sciences, Sargodha, Pakistan (No. RFMC/IRAB/Protocol/No.2025-005). All participants provided written informed consent before enrollment, participation was voluntary, no personally identifying information was recorded, and the data were analyzed in anonymized form. The study conformed to the principles of the Declaration of Helsinki.

3.1. Sample characteristics

A total of 483 adults with T2DM completed the interview. Their characteristics are summarized in Table 1. Only under two-thirds were men (64.4%), and the mean age was 44.8 ± 10.5 years, ranging from 18 to 76 years. Most were married (73.3%) and lived within a joint family system (75.2%). Formal education was limited: 42.7% had received no formal schooling, and the mean duration of formal education was 6.5 ± 6.2 years. The median monthly household income was PKR 51,000 (interquartile range 39,500–70,000). The mean duration of diagnosed T2DM was 7.2 ± 2.6 years; almost two-thirds reported a family history of diabetes (63.6%), and 41.0% were current cigarette smokers.

3.2. Self-management practices

Adherence to the individual practices is presented in Table 2 and Figure 1. Blood glucose monitoring was the most commonly reported behavior: 68.1% of the participants checked home blood sugars in line with physician advice, and 67.3% had a random blood glucose check at least once every three months. A well-balanced planned diet (66.5%), daily medication use (65.8%) and regular exercise (62.3%) were also reported by approximately two-thirds of the patients. Blood pressure monitoring (60.0%) and daily feet checking (58.0%) were somewhat less common. The two conspicuous gaps were the use of stress-management techniques, reported by fewer than half (45.5%), and three-monthly HbA1c testing, reported by only approximately one-quarter of patients (26.5%)—by a wide margin the least-practiced behavior in the cohort.

3.3. Practice-count index and its internal structure

Patients adhered to a mean of 5.20 ± 1.37 of the nine practices (median 5, interquartile range 4–6); the distribution is shown in Figure 2. Using the prespecified threshold, 342 patients (70.8%) achieved adequate self-management (≥ 5 of 9). Consistent with the broader literature, the nine practices largely behaved as independent behaviors rather than as facets of a single construct: internal consistency was low (Kuder–Richardson 20 = −0.11), the mean absolute interitem correlation was 0.08, and only one pair—daily feet checking and HbA1c testing—showed a moderate association (φ = 0.37; Figure 3). This pattern justified analyzing the practices individually and interpreting the count purely descriptively.

3.4. Factors associated with adequate self-management

Bivariate associations are shown in Table 3. Adequate self-management was significantly related to age, both as a continuous variable (44.0 ± 10.0 years in the adequate group versus 46.7 ± 11.3 years in the inadequate group; p = 0.011) and across age groups, decreasing from 75.2% among those aged 18–39 years to 61.6% among those aged 50 years or older (p = 0.042). It also varied by marital status (p = 0.035): married and widowed/separated/divorced patients were more likely to manage adequately (73.4% and 75.0%, respectively) than unmarried patients were (60.4%). A greater percentage of men than women achieved adequacy (74.0% versus 65.1%), a difference of borderline significance (p = 0.052). Education, income, family system, family history, disease duration and smoking showed no significant bivariate association with adequate self-management. At the level of individual practices, several behaviors differed by patient characteristics—for example, three-monthly random blood glucose checking differed by sex (χ² = 13.6, p < 0.001) and blood pressure monitoring by smoking status (χ² = 11.1, p = 0.001)—but these isolated associations should be read cautiously in light of multiple testing.

In the multivariable logistic regression (Table 4), two factors remained independently associated with adequate self-management. Each additional year of age reduced the odds of adequacy by approximately 4% (aOR 0.96, 95% CI 0.94–0.98; p < 0.001), and unmarried patients had markedly lower odds than married patients did (aOR 0.37, 95% CI 0.20–0.67; p = 0.001). The widowed/separated/divorced group did not differ significantly from the married reference group (aOR 1.87, 95% CI 0.72–4.83). Sex, education, income, disease duration, family system, family history and smoking status were not independent predictors. The model was statistically significant overall (likelihood ratio χ² = 28.8, p = 0.001), showed acceptable calibration (Hosmer–Lemeshow p = 0.086) and modest discrimination (AUC = 0.66), and explained a limited share of the variance (Nagelkerke R² = 0.083). A Poisson sensitivity model fitted to the practice count reproduced the age effect (incidence-rate ratio 0.995 per year, 95% CI 0.991–0.999; p = 0.020), supporting the robustness of the main finding.