Factors associated with chronic kidney disease stage among patients with uncontrolled diabetes and/or hypertension in Northeastern Thailand: a cross-sectional study

Abstract

Objective

To identify demographic, socioeconomic, behavioural and psychosocial factors associated with chronic kidney disease (CKD) stage among patients with uncontrolled diabetes mellitus and/or hypertension and CKD stages 1–3a.

Design

Cross-sectional study.

Setting

Two non-communicable disease clinics in Health Region 10, northeastern Thailand.

Participants

A total of 385 patients aged 30–70 years with CKD stages 1–3a and uncontrolled diabetes and/or hypertension.

Primary outcome measure

CKD stage at the time of assessment, obtained from clinical records, was dichotomised as Stage 1 versus Stages 2–3a. Multivariable logistic regression was used to identify factors associated with Stage 1. As this was a cross-sectional study, the outcome represented CKD stage at a single time point rather than CKD progression.

Results

Participants’ mean age was 59.50 years (SD 8.22), and 70.39% were female. In adjusted analyses, those aged 51–60 years had lower odds of Stage 1 than those aged ≥61 years (AOR=0.52; 95% CI 0.31 to 0.88; p=0.016). Compared with income >20 000 Baht, incomes <5000 Baht (AOR=5.49; 95% CI 1.35 to 22.29; p=0.017) and 10 001–15 000 Baht (AOR=6.94; 95% CI 1.55 to 31.06; p=0.011) were associated with higher odds of Stage 1. Low versus high medication-related self-care was also associated with higher odds of Stage 1 (AOR=2.15; 95% CI 1.30 to 3.56; p=0.003).

Conclusions

Age, income and medication-related self-care were associated with CKD stage. The unexpected associations of lower income and lower medication-related self-care with Stage 1 should be interpreted cautiously because of the cross-sectional design and potential reverse causation, selection bias and residual confounding. Longitudinal studies are needed to assess changes in kidney function over time.

STRENGTHS AND LIMITATIONS OF THIS STUDY.

• This study integrated demographic, socioeconomic, behavioural and psychosocial factors within a single analytical framework using a content-validated and reliability-tested questionnaire.

• The inclusion of patients with uncontrolled diabetes and/or hypertension in early-stage chronic kidney disease provides valuable insights into high-risk populations in primary care settings.

• The cross-sectional design precludes assessing changes in kidney function over time and limits temporal and causal inference.

• Self-reported measures of self-care behaviours may be subject to recall and social desirability biases.

• The study was conducted in a specific region of northeastern Thailand, which may limit the generalisability of the findings to other populations.

Introduction

Chronic kidney disease (CKD) is a major global public health concern. It affects over 700 million people worldwide and is among the leading causes of death.¹ The burden of CKD continues to rise, significantly reducing quality of life and increasing healthcare costs.² In Thailand, CKD affects about 17% of the population. The highest prevalence is in the northeastern region (26.8%), highlighting the urgent need for effective prevention and early intervention strategies.^{3 4}

According to Kidney Disease: Improving Global Outcomes guidelines, CKD is defined as abnormalities of kidney structure or function persisting for more than 3 months, as indicated by a decline in estimated glomerular filtration rate (eGFR) or markers of kidney damage.⁵ Early-stage CKD (stages 1–3a) represents a critical period for intervention, during which appropriate management can slow disease progression, reduce complications and delay the need for renal replacement therapy.^{6 7}

Uncontrolled diabetes mellitus and hypertension remain the leading causes of CKD and major contributors to disease progression. Persistent hyperglycaemia and elevated blood pressure damage glomerular structures, accelerating kidney function decline.⁸ To counter these effects, current management strategies emphasise comprehensive risk control. This includes glycaemic and blood pressure regulation, lipid management, a healthy diet, regular physical activity and adherence to medical treatment.⁹ Additionally, pharmacological interventions, particularly sodium-glucose cotransporter 2 inhibitors, have demonstrated effectiveness in reducing the progression of kidney disease and improving renal outcomes in patients with type 2 diabetes.¹⁰

Noncommunicable diseases (NCDs), including diabetes, hypertension and CKD, are major public health concerns in Thailand. CKD affects a substantial proportion of the population, with an estimated prevalence of approximately 17–18%.¹¹ The growing burden of NCDs is intensified by demographic changes, such as population ageing, and challenges in healthcare access, which together contribute to the rising prevalence. These demographic and systemic factors make it more likely that diabetes, hypertension and early-stage CKD will remain uncontrolled.

Beyond clinical and biological factors, psychosocial factors also influence chronic disease management. Health literacy and understanding of disease-related information are associated with the ability to perform appropriate self-care behaviours.¹² Evidence suggests that self-management support interventions can facilitate behavioural change and long-term adherence, especially in people with chronic conditions.¹³ Coping and adaptation to illness have also been associated with engagement in disease management.¹⁴ In CKD, these factors are associated with self-management behaviours, including glycaemic and blood pressure control.^{15 16}

Previous studies have examined factors associated with CKD stage and kidney outcomes; however, evidence focusing on patients with uncontrolled diabetes and/or hypertension and early-stage CKD in Thailand remains limited. The Thai healthcare system is organised into regional Health Regions that integrate services across provinces. Health Region 10 in northeastern Thailand—covering Ubon Ratchathani, Yasothon, Sisaket, Amnat Charoen and Mukdahan—has a high burden of CKD and related risk factors.³ This study therefore aimed to identify demographic, socioeconomic, behavioural and psychosocial factors associated with CKD stage among patients with uncontrolled diabetes and/or hypertension and CKD stages 1–3a in Health Region 10, Thailand.

Materials and methods

Study design and setting

This study reports Phase 1, the cross-sectional survey component of a three-phase research project. Phase 1 examined factors associated with CKD stage among patients with uncontrolled diabetes mellitus and/or hypertension. Data were collected between October 2025 and November 2025 at the NCD clinics of Warin Chamrap Hospital, Ubon Ratchathani Province, and Kut Chum Hospital, Yasothon Province, in Health Region 10, northeastern Thailand.

Study population, sample size and eligibility criteria

The study population comprised patients with uncontrolled diabetes mellitus and/or hypertension and CKD stages 1–3a receiving care at NCD clinics in Health Region 10. The sample size was calculated using Cochran’s formula (Z=1.96, p=0.50, e=0.05), yielding 384.16, which was rounded up to 385 participants. A multistage probability sampling procedure was employed. First, we randomly sampled Ubon Ratchathani and Yasothon provinces from the five provinces in Health Region 10 using simple random sampling. Second, we randomly sampled one district hospital in each sampled province: Warin Chamrap Hospital from the 23 district hospitals in Ubon Ratchathani Province and Kut Chum Hospital from the eight district hospitals in Yasothon Province. Third, eligible patients were randomly sampled without replacement from the patient databases of the two participating hospitals. The sampling frames comprised 1579 eligible patient records in the Health Data Centre (HDC) database of Warin Chamrap Hospital and 753 eligible patient records in the HOSxP hospital information system of Kut Chum Hospital. The final sample comprised 193 and 192 participants from the respective hospitals.

Inclusion criteria were age 30–70 years; CKD stages 1–3a (eGFR≥45 mL/min/1.73 m²); uncontrolled diabetes (glycated haemoglobin (HbA1c)≥7%) and/or uncontrolled hypertension (systolic blood pressure≥140 mmHg and/or diastolic blood pressure≥90 mmHg); having a family member or primary caregiver; and capacity to provide informed consent. Exclusion criteria were CKD stages 3b–5 (eGFR<45 mL/min/1.73 m²), an acute complication at recruitment, declining participation or withdrawal from the study.

Data collection instruments

Data were collected using a structured and self-administered questionnaire. Researchers developed this tool after reviewing relevant literature and consulting with experts. The instrument consisted of five sections.

1. The first section addressed demographic and clinical characteristics, including sex, age, marital status, education level, occupation, income, primary caregiver, food preparation, underlying diseases, duration of chronic conditions, duration of CKD and clinical parameters.

2. The second section focused on knowledge of CKD prevention. Knowledge was assessed using 10 items with three response options (true, false, uncertain). Correct responses were scored as 1, while incorrect and uncertain responses were scored as 0. Total scores were categorised as high, moderate or low using predefined cut-off points commonly used in health research.

3. The third section measured self-care behaviours. This was done using a 15-item, 4-point Likert scale, including both positively and negatively worded items with reverse scoring applied where appropriate. Mean scores were categorised as high, moderate or low based on standard scale-interpretation approaches.

4. The fourth section assessed self-efficacy. This section used a 15-item, 4-point Likert scale reflecting participants’ confidence in performing health-related behaviours. Scores were classified into three levels using the same criteria.

5. The fifth section addressed family support. Family support was measured using a 15-item, 4-point Likert scale assessing perceived emotional, informational and practical support from family members. Scores were categorised into three levels using the same approach.

Knowledge items were scored 1 for a correct response and 0 for an incorrect or uncertain response; total percentage scores were classified as high (≥80%), moderate (61%–79%) or low (≤60%). For the four-point scales, positively worded items were scored from 1 to 4 and negatively worded items were reverse-scored. Mean scores were classified according to the prespecified protocol as high (2.68–4.00), moderate (1.34–2.67) or low (1.00–1.33).^{17 18} The complete questionnaire is provided in online supplemental file 1.

Instrument development and quality assessment

The questionnaire was developed through a literature review, item development and expert consultation. Five experts—one nephrologist, one pharmacist, one nutritionist and two nurses—assessed content validity by rating each item’s relevance on a four-point scale. The content validity index values were 0.84 for knowledge, 0.93 for self-care behaviour, 0.87 for self-efficacy and 0.80 for family support, indicating acceptable to excellent content validity.

Reliability was evaluated in a pilot study of 30 patients with characteristics similar to the study population at the NCD clinic of Selaphum Hospital, Roi Et Province, outside the main study sites. The questionnaire was self-administered under the research team’s supervision.

The knowledge scale demonstrated acceptable reliability, with a Kuder-Richardson 20 (KR-20) coefficient of 0.82. The self-care behaviour, self-efficacy and family support scales also showed good internal consistency, with Cronbach’s alpha coefficients of 0.91, 0.86 and 0.83, respectively.

Variables and outcome measures

The primary outcome was CKD stage at the time of assessment, obtained from clinical records and dichotomised as Stage 1 versus Stages 2–3a. All participants had an established diagnosis of CKD recorded in their clinical records. Among these patients, CKD stage at the time of assessment was classified according to eGFR as Stage 1 (≥90 mL/min/1.73 m²), Stage 2 (60–89 mL/min/1.73 m²) or Stage 3a (45–59 mL/min/1.73 m²).⁵ This classification compared participants at Stage 1 with those at more advanced stages within early-stage CKD and was not used to assess disease progression. Independent variables included demographic factors (age, sex, education, occupation and income), clinical factors (duration of diabetes and/or hypertension, comorbidities and clinical parameters), behavioural factors (diet, medication use, exercise, stress management and healthcare-related self-care) and psychosocial factors (self-efficacy and family support). Questionnaire-derived variables were categorised using the prespecified scoring criteria. In the final model, low medication-related self-care was compared with high medication-related self-care as the reference category.

Data collection procedures

After randomly sampling the provinces and hospitals, hospital staff screened the HDC and HOSxP databases against the eligibility criteria. The investigators then randomly sampled eligible patients without replacement from each sampling frame. The investigators contacted the randomly sampled patients and invited them to attend an appointment for data collection. None of the invited patients declined participation. The final sample comprised 385 participants: 193 from Warin Chamrap Hospital in Ubon Ratchathani Province and 192 from Kut Chum Hospital in Yasothon Province.

The research team explained the study objectives, procedures, voluntary nature of participation and confidentiality protections. Written informed consent was obtained before enrolment, and no financial or non-financial incentives were offered.

Participants completed the questionnaires themselves, with assistance provided when needed. We obtained clinical data, including eGFR, blood pressure and laboratory results, from medical records. We checked completed questionnaires for completeness and accuracy before data entry and analysis.

This study did not collect visit frequency to primary-care practitioners or nephrology clinics. Consequently, we could not include healthcare-contact intensity in the analysis, which may have introduced residual confounding.

Statistical analysis

Data were analysed using IBM SPSS Statistics for Windows, V.26.0 (IBM Corp., Armonk, NY, USA). Continuous variables were summarised as mean (SD) or median (IQR), as appropriate, and categorical variables as n (%). Bivariate associations with CKD stage (Stage 1 vs Stages 2–3a) were assessed using Pearson’s χ² test or Fisher’s exact test when expected cell counts were <5. Variables with a bivariate p<0.25 and those specified a priori based on clinical relevance were entered into the initial multivariable logistic regression model. A purposeful-selection approach was then used to obtain the final model, following Bursac et al.¹⁹ The dependent variable was coded as 1 for Stage 1 and 0 for Stages 2–3a. Categorical predictors were dummy-coded using prespecified reference categories. Multicollinearity was assessed using variance inflation factors, and model fit was evaluated using the Hosmer-Lemeshow goodness-of-fit test. Adjusted odds ratios (AORs) with 95% CIs were reported. No data were missing, and all 385 participants were included in the analyses. Statistical significance was set at p<0.05.

Results

Participant characteristics

A total of 385 participants were included (table 1). Their mean age was 59.50 years (SD 8.22); 70.39% were female, 71.43% were married, 75.58% had completed primary education and 56.88% worked in agriculture. Most participants (90.65%) reported a family member as their primary caregiver. Diabetes alone was the most common underlying condition (46.49%), followed by hypertension alone (19.48%) and combined diabetes and hypertension (18.18%). Of the participants, 207 (53.77%) were classified as CKD Stage 1 and 178 (46.23%) as Stages 2–3a. The mean duration of the underlying condition and CKD was 8.75 and 2.46 years, respectively. Mean BMI was 25.34 kg/m², waist circumference was 86.53 cm, blood pressure was 135.51/77.06 mmHg, eGFR was 88.06 mL/min/1.73 m² and HbA1c was 7.93%.

Table 1. Demographic and clinical characteristics of participants (n=385).

Characteristics	Participants (n=385)
Age (years, mean±SD)	59.50±8.22
Income (Baht/month), median (IQR)	5000 (2000–10 000)
Sex, n (%)
Male	114 (29.61)
Female	271 (70.39)
Marital status, n (%)
Single	33 (8.57)
Married	275 (71.43)
Widowed/divorced/separated	77 (20.00)
Education level, n (%)
Primary education	291 (75.58)
Secondary education	86 (22.34)
Bachelor’s degree or higher	8 (2.08)
Occupation, n (%)
Agriculture	219 (56.88)
Non-agriculture	166 (43.12)
Primary caregiver, n (%)
Family member	349 (90.65)
Relative and others	36 (9.35)
Underlying diseases, n (%)
Diabetes mellitus only	179 (46.49)
Hypertension only	75 (19.48)
Diabetes mellitus and hypertension	70 (18.18)
Other comorbid combinations	61 (15.84)
Duration of underlying conditions (years, mean±SD)	8.75±5.46
CKD stage, n (%)
Stage 1	207 (53.77)
Stages 2–3a	178 (46.23)
Duration of CKD (years, mean±SD)	2.46±1.33
Clinical characteristics (mean±SD)
BMI (kg/m²)	25.34±4.56
Waist circumference (cm)	86.53±11.56
Systolic/diastolic blood pressure (mmHg)	135.51±14.80/77.06±12.00
Serum creatinine (mg/dL)	0.79±0.20
eGFR (mL/min/1.73 m²)	88.06±17.57
Fasting blood sugar (mg/dL)	145.62±51.51
HbA1c (%)	7.93±3.26

Continuous variables are presented as mean (SD) when normally distributed and as median (IQR) when skewed. Categorical variables are presented as n (%).

BMI, body mass index; CKD, chronic kidney disease; eGFR, estimated glomerular filtration rate; FBS, fasting blood sugar; HbA1c, glycated haemoglobin.

Knowledge, self-care behaviours, self-efficacy and family support

Participants’ knowledge levels varied across domains (figure 1a). High knowledge was most common for healthcare-related knowledge (79.5%), followed by medication-related knowledge (50.6%), whereas dietary knowledge was predominantly low (71.7%). Overall, 49.4% of participants had high knowledge and 43.4% had moderate knowledge.

Figure 1. Knowledge, self-care behaviours, self-efficacy and family support among participants. (a) Knowledge levels by domain and overall distribution. (b) Mean scores of self-care behaviours across domains. (c) Mean self-efficacy scores across domains. (d) Mean family support scores across domains. CKD, chronic kidney disease.

Self-care behaviour scores varied across domains (figure 1b). Healthcare-related self-care had the highest mean score (3.31), followed by exercise-related self-care (3.10), whereas dietary self-care had the lowest mean score (2.66).

Self-efficacy was high in most domains (figure 1c). Stress-management self-efficacy had the highest mean score (3.13), while dietary self-efficacy had the lowest (2.67).

Family support was high across all domains (figure 1d). Care-related support had the highest mean score (3.04), followed by emotional support (2.96) and informational support (2.92).

Bivariate analysis of factors associated with CKD stage

In bivariate analysis (table 2), age was significantly associated with CKD stage (χ²=15.00, df=3, p=0.002). No statistically significant associations were observed for occupation, monthly income, underlying diseases, diet- or medication-related knowledge, overall or domain-specific self-care, overall or diet-related self-efficacy, or emotional family support (all p>0.05), although underlying diseases approached statistical significance (p=0.052).

Table 2. Factors associated with CKD stage: bivariate analysis.

Variable	χ²	df	P value
Age	15.00	3	0.002
Occupation	8.44	5	0.133
Monthly income	8.09	4	0.088
Underlying diseases	10.98	5	0.052
Knowledge of diet	1.65	1	0.198
Knowledge of medication use	4.23	2	0.120
Self-care behaviours (overall)	1.84	1	0.175
Self-care behaviours related to diet	4.16	2	0.125
Self-care behaviours related to medication use	3.20	1	0.073
Self-efficacy (overall)	3.17	1	0.075
Self-efficacy related to diet	4.16	2	0.125
Family support: emotional	1.46	1	0.226

df, degrees of freedom; χ², chi-square statistic.

Multivariable analysis of factors associated with CKD stage

In multivariable logistic regression (table 3), participants aged 51–60 years had lower odds of CKD Stage 1 versus Stages 2–3a than those aged ≥61 years (AOR=0.52, 95% CI 0.31 to 0.88; p=0.016).

Table 3. Multivariable logistic regression analysis of factors associated with CKD Stage 1 versus Stages 2–3a.

Variables	B	SE	Wald	P value	AOR	95% CI
Age 51–60 years (Ref: ≥61 years)	−0.641	0.26	5.77	0.016	0.52	0.31–0.88
Income<5000 Baht (Ref: >20 000 Baht)	1.704	0.71	5.68	0.017	5.49	1.35–22.29
Income 10 001–15 000 Baht (Ref: >20 000 Baht)	1.937	0.76	6.41	0.011	6.94	1.55–31.06
Low medication-related self-care (Ref: high)	0.767	0.25	8.90	0.003	2.15	1.30–3.56

The dependent variable was coded 1 (CKD Stage 1) and 0 (Stages 2–3a). Candidate predictors with bivariate p<0.25 and variables selected a priori for clinical importance were considered for the multivariable model. Reference categories are shown in parentheses. Questionnaire variables were categorised using the prespecified scoring rules described in the Methods. Multicollinearity was assessed using variance inflation factors, and calibration was assessed using the Hosmer-Lemeshow test. No data were missing; all 385 participants were included in the multivariable analysis. Statistical significance was set at p<0.05. The maximum VIF was 5.24. The Hosmer-Lemeshow goodness-of-fit test yielded χ²=6.99 (df=8, p=0.537).

AOR, adjusted odds ratio; B, regression coefficient; SE, standard error; Wald, Wald chi-square statistic.

Participants with monthly incomes <5000 Baht (AOR=5.49, 95% CI 1.35 to 22.29; p=0.017) or 10 001–15 000 Baht (AOR=6.94, 95% CI 1.55 to 31.06; p=0.011) had higher odds of Stage 1 than those earning >20 000 Baht.

Participants with low medication-related self-care had higher odds of Stage 1 than those with high medication-related self-care (AOR=2.15, 95% CI 1.30 to 3.56; p=0.003).

Discussion

This cross-sectional study identified demographic, socioeconomic and behavioural factors associated with CKD stage among patients with uncontrolled diabetes mellitus and/or hypertension and early-stage CKD. The findings describe differences between participants classified as Stage 1 and those classified as Stages 2–3a at a single assessment; they do not describe change in kidney function over time.

The participants represented a high-risk clinic population, with a predominance of older adults and people with low educational attainment. Age-related structural and functional kidney changes are well established,^5–7 while socioeconomic constraints may be associated with access to health information and opportunities for self-management.²⁰ The multinational REVEAL-CKD study also reported that many people with early-stage CKD remain undiagnosed, particularly in high-risk populations.²¹ These contextual patterns are consistent with the substantial CKD burden reported in Thailand and other settings.^{3 11}

Clinical findings highlighted metabolic risks. The mean BMI was in the overweight range, and the mean waist circumference indicated central adiposity, while mean fasting blood glucose and HbA1c showed suboptimal glycaemic control. These characteristics are clinically relevant to kidney health,^22–24 but their cross-sectional measurement does not establish their temporal relationship with CKD stage.

Overall knowledge levels were moderate to high; however, substantial gaps in dietary knowledge persisted, consistent with previous evidence that CKD-related education is complex and not fully understood by patients.²⁵ While knowledge of medication use and general healthcare was relatively high—possibly reflecting exposure to routine clinical care, although visit frequency was not measured—understanding of dietary management remained limited. This contrast indicates that, although patients are more familiar with medication and general care, the complexity of dietary concepts remains a significant challenge. Consequently, limited understanding of dietary management may hinder effective long-term disease control, underscoring the need for continuous education and patient engagement, as emphasised in recent approaches to CKD care and patient empowerment.²⁶

Despite these knowledge gaps, participants reported high self-care in most domains, whereas dietary self-care was moderate and had the lowest mean score. This pattern may be associated with healthcare support and routine follow-up in NCD clinics, although visit frequency was not measured.^{9 27} Previous studies have shown that better self-management practices are associated with improved clinical outcomes and slower disease progression among patients with CKD.²⁷ The comparatively low dietary self-care score may reflect cultural and socioeconomic barriers to modifying long-term dietary behaviour.²⁸

Self-efficacy was high in most domains, particularly stress management, whereas dietary self-efficacy was moderate. These findings indicate that participants generally felt confident in performing health-related behaviours but were less confident in dietary management. Prior studies have linked higher self-efficacy with better CKD self-management.¹⁵ A trial protocol is evaluating whether digital self-management support can help patients manage diabetic kidney disease.²⁹ Patient-centred and broader self-care interventions may also support long-term self-care in chronic illness.^{30 31} These findings highlight the need for tailored dietary interventions that address barriers and strengthen patients’ confidence in dietary self-management.

Family support was also high across all domains, particularly in caregiving. Families assisted with daily routines, offered emotional encouragement and helped patients adhere to medical regimens. These findings underscore the important role of social support in chronic disease management. Additionally, evidence indicates positive associations between social support and both self-management behaviours and treatment adherence among patients with CKD.^32–34 Moreover, social support may enhance patients’ self-efficacy and engagement in self-care activities, thereby facilitating better management of chronic conditions.^{33 34}

The multivariable analysis identified associations between CKD stage and age, income and medication-related self-care. Participants aged 51–60 years had lower odds of Stage 1 than those aged ≥61 years. Because age and kidney function were assessed concurrently in clinic patients with early-stage CKD, this finding represents a cross-sectional association rather than an age-related trajectory of kidney decline.

Lower income was unexpectedly associated with higher odds of Stage 1. This finding does not indicate that lower income improves kidney outcomes and may reflect differences in clinic attendance, diagnosis, access to testing, study selection, unmeasured clinical factors or reverse causation. The wide CIs also indicate imprecision. Longitudinal studies with repeated eGFR and albuminuria measurements are needed to clarify the temporal relationship between socioeconomic conditions and kidney function.

Lower medication-related self-care was also unexpectedly associated with higher odds of Stage 1. Because Stage 1 was coded as the outcome and high medication-related self-care was the reference category, the observed AOR should not be interpreted as evidence that poorer self-care protects against CKD progression. Patients with more advanced CKD may have intensified their medication routines following diagnosis, resulting in reverse causation. Self-reported behaviour may also have introduced social desirability or misclassification bias. The cross-sectional design cannot establish whether medication-related self-care preceded CKD stage.

Overall, these findings identify patient characteristics associated with CKD stage within the sampled clinics but not determinants of CKD progression. Because CKD assessment and risk stratification involve both eGFR and markers of kidney damage, particularly albuminuria, future longitudinal studies should include repeated measurements of these parameters.³⁵ Early recognition and comprehensive management of CKD, including blood pressure and glycaemic control, appropriate pharmacotherapy and dietary management, remain essential.³⁶

Limitations

This study has several limitations. The cross-sectional design precluded assessment of temporal relationships and CKD progression, prevented causal inference and was susceptible to reverse causation. Despite multistage probability sampling, inclusion of only two provinces and two hospitals may have introduced selection bias and limited generalisability. Self-reported measures were susceptible to recall, social desirability and misclassification biases. Residual confounding may also remain because relevant clinical and healthcare-related factors, including the frequency of primary-care or nephrology visits, were not measured or available.

Conclusions

Among patients with uncontrolled diabetes and/or hypertension and CKD stages 1–3a, age, income and medication-related self-care were associated with CKD stage at the time of assessment. Lower income and lower medication-related self-care were unexpectedly associated with higher odds of Stage 1. These findings may reflect reverse causation, selection bias, measurement error or residual confounding and should not be interpreted causally. Prospective studies with repeated kidney function measurements are needed to assess CKD progression and its temporal predictors.

Data availability statement

Data are available upon reasonable request.