diabetes research and clinical practice 107 (2015) 148–156

Contents available at ScienceDirect

Diabetes Research and Clinical Practice jou rnal hom ep ag e: w ww.e l s e v i er . c om/ loca te / d i ab r es

Uncontrolled diabetes mellitus: Prevalence and risk factors among people with type 2 diabetes mellitus in an Urban District of Karachi, Pakistan Fahad Javaid Siddiqui a,b,*, Bilal Iqbal Avan c, Sadia Mahmud d,e, Debra J. Nanan f,g, Abdul Jabbar h, Pryseley Nkouibert Assam a,b a

Center for Quantitative Medicine, Office of Clinical Sciences, Duke-NUS Graduate Medical School Singapore, Singapore b Singapore Clinical Research Institute, Singapore c London School of Hygiene & Tropical Medicine, London, UK d Department of Medicine, Aga Khan University, Pakistan e Department of Paediatrics & Child Health, Aga Khan University, Pakistan f Pacific Health & Development Sciences, University of Victoria, Victoria, BC, Canada g School of Public Health and Social Policy, University of Victoria, Victoria, BC, Canada h The Aga Khan University, Karachi, Pakistan

article info

abstract

Article history:

Aims: This study aimed to explore the prevalence of, and factors associated with, uncon-

Received 20 April 2014

trolled diabetes mellitus (UDM) in a community setting in Pakistan.

Received in revised form

Methodology: A single-center, cross-sectional study, conducted in a community-based spe-

11 August 2014

cialized care center (SCC) for diabetes in District Central Karachi, in 2003, registered 452 type

Accepted 15 September 2014

2 DM participants, tested for HbA1c and interviewed face-to-face for other information.

Available online 5 October 2014

Logistic regression analysis was conducted to identify factors associated with UDM. Results: Prevalence of UDM among diabetes patients was found to be 38.9% (95% CI: 34.4–

Keywords:

43.4%). Multivariable logistic regression model analysis indicated that age 8.4 59.0

a Authors categorized as: normal = 7.5. We considered categories ‘normal’ and ‘good’ as the same. HbA1c% = NGSP units (IFCC units): 6.5 = 48; 7 = 7.4 51.6

150

2.3.

diabetes research and clinical practice 107 (2015) 148–156

Study conduct and assessments

During the clinic consultation, treating physicians identified potentially eligible patients using a check list and referred them to a study coordinator onsite for their informed consent. After obtaining consent, a face-to-face interview was conducted, using a validated structured questionnaire. Afterwards patient’s height and weight measurements were taken, and a blood specimen was collected. During the interview, information was collected on: demographics, socioeconomics, diabetes knowledge, treatment modalities used, lifestyle measures adopted and expenditure on diabetes treatment. Socioeconomic status was evaluated by assessing per capita household income. Literacy was defined as having any formal schooling. The number of cups of tea or coffee drank per day, as well as the type and quantity of sweetener used, were recorded. Diabetes knowledge was assessed using 8 items, eliciting awareness about sources of information, role of insulin, non-suitable components of diet (using local terms) for PWD, exercise, complications, and treatment modalities. Items were similar to those used in the Michigan Diabetes Knowledge Test (MDKT) [20]. However, anticipating language and cultural differences, it was pretested on 5% of the calculated sample size and customized. Correct response for each of the 8 items was assigned an item-score of 1, and 0 otherwise. A diabetes knowledge score was then computed as the sum of the 8 item-scores, ranging from 0 to 8. Anxiety & depression was assessed using Aga Khan University Anxiety Depression Scale AKUADS [21]. Weight was measured in light clothing. Venous blood was tested for glycated hemoglobin through ion exchange resin method (Pre-Fil1, Stanbio Glycohemoglobin; Procedure No. P350; Boerne, TX, USA). UDM was defined as glycated hemoglobin levels of 8% (64 mmol/mol) or above.

2.4.

mean and standard deviation, median and interquartile range, as appropriate, of demographic and socioeconomic characteristics, treatment modalities, lifestyle measures and diabetes knowledge by UDM status. Univariable and multivariable logistic regression models were used to estimate the unadjusted and adjusted OR between potential risk factors and UDM respectively [25,26]. First, univariable logistic regression models were fitted for each potential risk factor. Second, stepwise, forward and backward automated variable selection methods were used to identify independent risk factors [29–31]. The significance level for a risk factor to enter or stay in the model was set at 0.20; p-value to remove was set at 0.25. Third, variables selected by any of the automated variable selection methods were included in a multivariable logistic regression model and a backward elimination procedure, using likelihood ratio tests, was performed based on a significance level of 0.05. Biologically plausible interactions between independent risk factors were also evaluated. The Hosmer and Lemeshow test was used to evaluate the goodness of fit of the multivariable model [32]. Significance level was set at 5% unless stated otherwise. SAS Version 9.2 (SAS Institute, Cary, NC) was used for the analysis.

3.

Four hundred and fifty six (456) eligible patients identified during the recruitment period of the study gave their consent and were subsequently enrolled. Blood specimens of 4 patients could not be analyzed for HbA1c due to insufficient serum (n = 3) or broken tube (n = 1). The analyzed data contained 452 (99.1%) diabetes patients, with a median duration of diabetes of 5.5 (0.5–35.5) years.

Sample size 3.1.

A sample size of 383 diabetes patients was calculated based on the following assumptions – an expected prevalence of UDM of 50% among diabetes patients; a 5% margin of error; a confidence level of 95%; and a population of diabetes patients above 25 years of 100,000. A 50% prevalence of UDM, HbA1c  8, was chosen based on the results of previous studies (range: 42–67%; SD: 6.25) [22–24] (and 50% prevalence gives the largest sample size). Regarding identification of risk factors, a conservative prevalence of UDM of 42% was considered, yielding an expected 160 UDM cases out of the planned 383 patients. Based on the rule of thumb recommending 10 events per associated factor, 160 UDM cases projected at most 16 factors. Therefore 383 patients were sufficient for both study objectives, although 456 individuals consented to participate during the study period.

2.5.

Results

Statistical analysis

The proportion of UDM, and its corresponding 95% confidence limits (CL), were used to estimate the prevalence of UDM. The 95% CL were calculated based on the normal approximation method [25,26] and the Wilson score method [27,28]. Summary tables were used to describe the frequency and proportion,

Patient characteristics

The characteristics of the study sample are described in Table 2. The median age of the analyzed study sample was 50 (range: 26–88) years. Two thirds (72.1%) were women and the mean BMI was 27.1 kg/m2 (SD: 4.6) and 62.8% of the patients were obese (BMI  30 kg/m2). Approximately 66.2% were literate with a median years of schooling of 8.0 (range: 0.0–18.0) years. Only a fifth (21.7%) of the patients were gainfully employed and the median monthly income was rupees 6500 (IQR: 4000–10,000), equivalent to approximately $112 (financial year 2002–2003). The median household size was 6 (1–30) persons and 75.2% of the participants were currently married. Almost all participants (94.3%) were tea drinkers with 44 (10.3%) of those adding natural sweetener (white sugar) to their drink. The amount used did not exceed 2 teaspoonful per serving with only 12 PWD adding 1 or 2 teaspoonful (Table 3).

3.2.

UDM prevalence

The prevalence of UDM as per clinical practice during the conduct of the study, defined as HbA1c level of 8% [64 mmol/ mol], was 38.9% (95% CL: 33.3%, 44.8%) among people with

151

diabetes research and clinical practice 107 (2015) 148–156

Table 2 – Characteristics of people with type 2 diabetes mellitus attending specialized care center in Karachi, Pakistan (2002–2003). Variable

Socio-demographic characteristics Age group, n (%) 7% [53 mmol/mol]) is 74% (95% CI: 70.1%, 78.2%), approximately two times the prevalence based on the former cut-off.

Table 3 – Unadjusted odds ratios (OR) from logistic regression for factors associated with uncontrolled diabetes mellitus among persons with type 2 diabetes mellitus attending specialized care center in Karachi, Pakistan (2002–2003). Comparison

Characteristic

Univariable OR and 95% CL

p-Value

Age group Gender Education

2 2 vs. >2

1.26 (0.63; 2.53) 1.66 (0.84; 3.29)

0.21

No of consultations in last 3 months

Less than once a month vs. once a month More than once a month vs. once a month

1.55 (1.03; 2.35) 1.59 (0.89; 2.84)

0.08

No of blood tests in last 3 months

Less than once a month vs. once a month More than once a month vs. once a month

1.38 (0.91; 2.09) 1.33 (0.77; 2.30)

0.28

Body mass index

Overweight vs. obese Under-normal vs. obese Anxiousjdepressed vs. stable

0.92 0.97 1.29 1.02 1.25 1.33 1.23 2.53 1.22 1.22

0.95

Anxiety and depression DM knowledge score Sources of diabetes information Do any physical activity Consume oily and or sweet food Drink tea/coffee Number of tea/coffee cups Tea or coffee sweetener

Hospital vs. clinic

Only doctorjnurse vs. multiple sources No vs. yes Yes vs. no Yes vs. no Yes vs. no

(0.63; (0.91; (0.79; (1.18; (0.98; (0.78; (1.00;

(0.56; (0.58; (0.88; (0.91; (0.84; (0.91; (0.80; (0.93; (1.02; (0.80;

1.51) 1.62) 1.89) 1.15) 1.86) 1.95) 1.91) 6.90) 1.45) 1.87)

0.2 0.69 0.27 0.15 0.35 0.07 0.03 0.35

diabetes research and clinical practice 107 (2015) 148–156

3.4.

Fig. 1 – Distribution of HbA1c levels among persons with type 2 diabetes mellitus attending specialized care center in Karachi, Pakistan (2002–2003).

Multivariable association

A multivariable logistic regression analysis indicated that age 8%. The previously published studies reported quite varied estimates, mostly owing to the differences in cut-offs [6,14–18]. Even

Fig. 2 – Adjusted odds ratios from multivariable regression model for factors associated with uncontrolled diabetes mellitus among persons with type 2 diabetes attending specialized care center in Karachi, Pakistan (2002–2003).

154

diabetes research and clinical practice 107 (2015) 148–156

so burden and management of UDM is not uniform across healthcare settings. Based on the estimates reported in this and in previous studies conducted in Pakistan, an impending epidemic of DM complications is feared as most patients attending tertiary care hospitals had UDM. Khan et al. despite using a ‘high’ cutoff of 8.5% (69 mmol/mol) found a UDM prevalence of 59%, from 3000 blood samples received in a clinical laboratory of a tertiary care hospital. Also, Basit et al. reported 81% of 2200 patients had UDM using the current cut-off of 7% (53 mmol/ mol) among patients attending a tertiary care hospital. In our study, patients younger than 50 years of age were more prone to have UDM, supporting prior published findings that younger people with type 2 diabetes are more likely to have UDM [24,34,35]. The factors underlying the elevated HbA1c levels among younger adults are not fully understood. Younger adults may have difficulty attending clinic appointments and assuming self-care activities, given their greater work and family responsibilities; they may also be more affected by rapid changes in lifestyle as reflected by obesity pattern, an indicator of dietary pattern and physical activity [36]. Previous studies have shown older people are generally more compliant [36]. Our results did not indicate any significant association between UDM and BMI, as well as physical activity – which may vary by age groups. However, our analysis did not indicate a significant interaction between age group and physical activity. Nevertheless, currently younger PWD may require more attention or monitoring. In our study, PWD diagnosed at hospitals, as compared to SCC, were more likely to have UDM. It is not entirely clear how to explain this association. However, in the out-of-pocket healthcare settings in Pakistan, patients who access hospitals are the ones who are either able to pay or are acutely sick. Our study population was mostly of low SES. When such patients are diagnosed in hospitals it is likely that they might be experiencing relatively severe symptoms at the time of diagnosis, a possible indicator of sub-optimal health seeking behavior. Some of them with poor health seeking behavior may not improve their attention toward health even after diagnosis. Thus such patients are likely to have uncontrolled DM. If so, the place of diagnosis may make substantial impact on patients’ behavior due to confidence, trust and better understanding of the disease at the time of diagnosis. We could not evaluate these postulations owing to the crosssectional nature of our study, and recommend further investigation based on prospective studies on incident cases. PWD who got information through multiple sources had better glycemic control as compared to those who only received information from healthcare staff. Continued reinforcement is important to maintain motivation of PWD so that they continue following medical advice [37]. In district Karachi Central 78% men and 74% women are literate, 82% had access to television and 58% of households are exposed to newspaper [38]. These statistics support our finding that access to multiple sources of information existed. We also found that those with multiple sources had higher DM knowledge score and more years of schooling ( p-values:

Uncontrolled diabetes mellitus: prevalence and risk factors among people with type 2 diabetes mellitus in an Urban District of Karachi, Pakistan.

This study aimed to explore the prevalence of, and factors associated with, uncontrolled diabetes mellitus (UDM) in a community setting in Pakistan...
728KB Sizes 0 Downloads 5 Views