This is an open-access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in JMIR Research Protocols, is properly cited. The complete bibliographic information, a link to the original publication on https://www.researchprotocols.org, as well as this copyright and license information must be included.
Background
Yearly screening for microvascular complications of diabetes mellitus (DM), namely diabetic retinopathy (DR), diabetic nephropathy (DN), and diabetic foot complications (DFC), is recommended to reduce their incidence, and delay or prevent their progression. Poor adherence to screening is common, but prospective data on the magnitude and predictors of nonadherence to DR, DN, and DFC screening in Singapore are unavailable.
Objective
The Understanding Non-Adherence to Diabetes Complications Screening study aims to determine the rates, predictors, and clinical and economic impact of nonadherence to diabetic complications screening in patients with type 2 diabetes in Singapore. The study describes the methodology and participants’ baseline characteristics that may be associated with nonadherence to DM complications screening.
Methods
In this prospective, mixed methods, clinic-based study, patients who underwent DR, DN, or DFC screening and were offered an annual rescreening appointment, were recruited from 6 primary care centers. Patients’ sociodemographic, medical, clinical, and patient-reported characteristics were recorded at baseline. Nonadherence to DR, DN, or DFC screening is defined as not attending the annual rescreening appointment within 4 months of the scheduled rescreening date. Adherence and clinical data will be recorded at 16 months post enrollment. Additionally, selected participants and health care professionals will undergo qualitative interviews to elicit barriers or facilitators of adherence to rescreening.
Results
Ethical approval was obtained in November 2016. Study enrollment commenced across the 6 sites between June 2018 and February 2019, and baseline data collection ended at all sites in January 2020. 974 eligible patients (2123 screenings; median age of 61.0, IQR 55.0-67.0 years; male: 515, 52.9%; Chinese: 624, 64.1%) consented and completed the baseline assessment. Of these, 734 (75.4%), 603 (61.9%), and 786 (80.7%) attended DR, DN, and DFC screening, respectively. Most (n=793, 81.4%) attended more than 1 complication screening on the same day; had received secondary or lower education (n=701, 71.9%); had hypertension (n=711, 73.4%) and dyslipidemia (n=828, 85.1%); and 43.1% (n=419) were obese (BMI>27.5 kg/m2). Median DM duration and hemoglobin A1c levels were 6.3 (IQR 3.0-12.0) years and 6.9% (6.4%-7.6%), respectively. Over half (n=532, 55.1%) had not received prior DM education. Furthermore, participants reported low levels of diabetes-related self-efficacy (median 1.4, IQR 1.0-3.9 out of 5).
Conclusions
At baseline, we have successfully enrolled almost 1000 patients with type 2 diabetes scheduled for annual DR, DN, or DFC rescreening, and potential predictors of nonadherence to rescreening were systematically collected. Follow-up phases will focus on establishing the rates and associated modifiable predictors of nonadherence to DR, DN, or DFC rescreening, which may inform program initiatives.
International Registered Report Identifier (IRRID)
Microvascular complications of diabetes mellitus (DM), namely diabetic retinopathy (DR), diabetic nephropathy (DN), and diabetic foot complications (DFC) are common, and impose a profound quality of life, health, and economic burden on the individual and society [1,2]. Regular screening for these complications can reduce their incidence and delay or prevent the progression of blindness, end-stage renal disease, and diabetic foot ulcers [3-7]. In Singapore, guidelines recommend that individuals with DM have a dilated fundus examination, assessment of urine albumin excretion and serum creatinine, and foot examination or education annually. However, screening programs are only cost- and clinically-effective if uptake is high, and studies conducted overseas have shown poor adherence to screening, ranging from 20% to 60% [8-10]. At a population-based level, the Singapore Epidemiology of Eye Disease study (2004-2011) showed that 80% of individuals with DM had undiagnosed DR, a proxy indicator of nonadherence to DR screening [11]. However, current data on nonadherence to DR, DN, and DFC screening in Singapore are lacking.
Several factors have been associated with nonadherence to DM complications screening including low educational attainment, issues with appointment access, and lack of social or family support [8,12-17]. However, most existing research is limited by a retrospective study design, or only using qualitative data collection, meaning that predictors of nonadherence to DM complications screening remain largely unknown. Additionally, nonadherence to screening has been associated with poor DM control [6,18], more severe disease [19,20], frequent treatment [21], and hospitalization needs [6]. Indeed, a systematic review exploring patients’ nonattendance at eye screening programs revealed that repeated nonattendance was associated with an increased risk of sight-threatening DR [20]. However, the studies included were conducted mostly in Western populations and may not be applicable to Asian populations due to different screening guidelines, and regional, political, social, economic, cultural, and health care systems.
To date, there are no prospective and contemporary population data on the rates; predictors; and clinical and economic impact of nonadherence to DM complications screening in Singapore. Furthermore, our understanding of the barriers or facilitators to screening adherence faced by patients with DM is limited. Such information is crucial for health care professionals and policy makers to make decisions on resource allocation and for informing interventions to improve adherence to DM screening programs. Against this background, we are conducting a prospective, multicentered clinic-based study (Understanding Non-Adherence to Diabetes Complications Screening [UNADS]) to ascertain the rates (aim 1); associated predictors (aims 2a and 2b); and clinical (aim 3) and economic impact (aim 4) of nonadherence to DM complications screening in primary care “polyclinic” patients with type 2 DM (T2DM) in Singapore using a mixed methods approach. In this paper, we provide a detailed description of the study design, data collection tools and protocols, and associated analytical plans. Furthermore, we present the patients’ baseline characteristics and suggest factors that may potentially be associated with nonadherence to DM complications screening.
MethodsConceptual Framework
We used Green and Kreuter Predisposing, Reinforcing, and Enabling Constructs in Educational Diagnosis and Evaluation (PRECEDE) and Policy, Regulatory, and Organizational Constructs in Educational and Environmental Development (PROCEED) model [22] to inform our mixed methods study design. This model has two distinct components: (1) “PRECEDE” which constitutes five diagnostic and planning phases to facilitate the identification of priorities and setting of objectives, and (2) PROCEED which consists of four phases and focuses on the identification of criteria for policy implementation and subsequent evaluation. PRECEDE-PROCEED has been widely used to identify and conceptualize barriers to health care screening in other health fields [23-26]. Our study used the first 5 “PRECEDE” phases (Figure S1 in Multimedia Appendix 1). The first phase (social diagnosis) establishes the health problems which impact the quality of life of the target population and the second phase (epidemiological diagnosis) documents the importance of these problems through survey and epidemiological data. The third phase (behavioral and environmental diagnosis) focuses on the systematic identification of health practices and other factors linked to health problems. The fourth phase (educational and ecological diagnosis) identifies the factors that, when modified, will most likely result in the desired behavior change. These factors are then classified as predisposing, enabling, or reinforcing. The last phase (administrative or policy assessment) of PRECEDE analyses policies, resources, and circumstances prevailing organizational situations that could hinder or facilitate meeting the objectives of the fourth phase. Based on our consultations with senior health care professionals from the SingHealth Polyclinics (SHP) and 2 National Healthcare Group Polyclinics (NHGP), and data from the NHG DM registry and Singapore Epidemiology of Eye Disease study, nonadherence to DM complications screening was already identified as a major public health issue in Singapore [11,27]. Therefore, in this study, we focus on the implementation of phases 2-5 via a prospective, mixed methods study that combines a quantitative data collection phase followed by in-depth qualitative interviews.
Study Setting and Design
In Singapore, DM complications screening is performed mainly at 23 primary health care government polyclinics. Of these, 4 SHP and 2 NHGPs were involved in our study, based on high attendance by minority ethnic groups, that is, Malays (Geylang [NHGP]) and Indians (Hougang [NHGP]), and high patient volume to maximize recruitment potential (ie, Bedok, Bukit Merah, Outram, and Pasir Ris [SHP]).
Standardized and uniform research protocols were applied at all sites. Each participant completed a baseline assessment on the day of enrollment and will be followed up at 16 months post enrollment (Figure 1). Baseline assessments involved the collection of quantitative data including clinical and sociodemographic information, and patient-reported outcomes. At 16 months following the baseline assessment (ie, up to 4 months after the scheduled annual rescreen appointment), a trained clinical research coordinator will determine if the participant attended the annual rescreening appointment from polyclinic records. Participants who attended or did not attend rescreening will be deemed “adherent” or “nonadherent,” respectively. The clinical research coordinator will then arrange a follow-up visit for all participants (both adherent and nonadherent) at the study site to collect study-related clinical information that is not available in the electronic medical records (EMRs). Furthermore, purposively selected participants will be invited to participate in focus groups or semistructured interviews regarding barriers or facilitators to adherence with DM complications screening. In addition, semistructured interviews will also be conducted with health care professionals (eg, polyclinic doctors and allied health professionals) to elicit their perceptions on why some patients do not comply with their screening visits.
The study was approved by the SingHealth Centralized institutional review board (reference #2016/3041), and all patients enrolled in this study provided written informed consent prior to participation. All health care professionals who will be enrolled in qualitative interviews will be verbally consented prior to participation. The study was conducted in accordance with the Declaration of Helsinki.
To ensure confidentiality, each patient is assigned a unique identification number linked to the identifiable data (eg, name, contact details), which are saved separately on a password-protected system in a password-protected file, at each study site. Hard copies of the data were stored in secured cabinets at each study site. Audio files will be coded with the participant’s identification number but without any information that could be used to identify them. Audio files and transcripts were stored in password-protected folders in shared drives of the Singapore Eye Research Institute. All data collected, including transcripts of the audio files, will be only accessible to approved research team members.
Each study participant will receive a reimbursement of S $20 (US $15.29) for each screening type for which they are recruited as part of the study. For example, a participant who completes baseline assessments for attending DR and DN screening will receive a total of S $40 (US $30.57) worth of vouchers. Similarly, the participant will receive a total of S $40 (US $30.57) worth of vouchers upon completing the respective follow-up assessments.
Study Population and ProceduresQuantitative PhasePopulation
Patients with T2DM who underwent DM complications screening at one of the study sites and received a referral for an annual rescreen at the participating sites were eligible. All eligible participants were Singaporean citizens or permanent residents of Chinese, Malay, Indian, or Eurasian ethnicity; aged 21 years and older; free of significant hearing impairment that could affect the study protocol, and cognitive impairment assessed using the 6-item Cognitive Impairment Test [28]. Patients who had type 1 DM or with complications necessitating referral to tertiary care (eg, presence of moderate or worse DR, overt proteinuria, or neuropathy) were deemed ineligible to participate. Each patient was assigned a unique study identification number for each type of screening attended and enrolled in this study. For example, a patient attending DR and DN screening appointments on the same or different days (bundle screening) was assigned 2 unique study identification numbers.
Sample Size Determination
Assuming the nonadherence rate for DR, DN, and DFC is 50%, 30%, and 40%, a sample size of 1111 for DR, 933 for DN, and 1067 for DFC was needed to obtain a 95% CI of ±3% around the rates of nonadherence estimates with 80% power. Accounting for nonresponse and loss to follow-up of 20%, a sample size of 1444 for DR, 1213 for DN, and 1387 for DFC was needed. Furthermore, assuming a nonadherence rate of 50% and a probability of exposure (barriers or facilitators) of 15% among those who are adherent, these proposed sample sizes would be able to detect an odds ratio (OR) of 1.47, 1.6, and 1.5 of nonadherence to DR, DN, and DFC screening, respectively, for a barrier or facilitator. Additionally, assuming that the effect of bundle screening would be 30% on DN and DFC screening where DR screening (n=1444) was treated as a constant, we required 849 and 971 for DN and DFC, respectively. In total, we required a sample of 3264 patients.
Data Collection
All study assessments were undertaken by trained project members. A full description of all study procedures is provided in Table S1 in Multimedia Appendix 2.
Sociodemographic, medical, and diabetes-related information and social history were collected via an in-house questionnaire and from EMR.
Anthropometric parameters: height and weight were measured on the day of assessment if not available in the participant’s EMR.
Clinical parameters: blood pressure (BP) was assessed using a digital automatic BP monitor [29] Dinamap model Pro Series DP110X-RW, 100V2; GE Medical Systems Information Technologies, Inc) on site if not available in EMR. Participants’ hemoglobin A1c (HbA1c), lipids (low- or high-density lipoprotein cholesterol, triglycerides, total cholesterol), serum creatinine, urine albumin creatinine ratio, and estimated glomerular filtration rate (eGFR) were collected from EMR if assessed 12 months or less. Otherwise, a venous blood sample drawn via venipuncture and a midstream urine sample were sent to Innoquest Laboratories Pte Ltd or National Healthcare Diagnostics, in Singapore for same-day analysis. The eGFR was calculated using the Chronic Kidney Disease Epidemiology Collaboration formula [30].
Ocular parameters: retinal graded results of participants who attended the Singapore Integrated Diabetic Retinopathy Programme, a teleophthalmology-based screening program, were obtained from participants’ EMR.
Patient-reported outcomes hypothesized to be associated with screening adherence were assessed using the following validated questionnaires: Revised Diabetes Knowledge Test [31,32], Health Literacy Test for Singapore questionnaire [33], 3-item Health Literacy Questionnaire [34], Diabetes Empowerment Scale (Short form)–Self Efficacy [35], Problem Areas in Diabetes Questionnaire [36], and Questions on Social Support [37,38].
Specially crafted items designed to elicit barriers from the patient’s perspective based on known barriers from the literature and expert panel input were also included (eg, history of adherence or nonadherence to DM complications screening, time taken to reach screening appointments, DM-related education needs and experiences, and attitudes toward diabetes control).
All baseline quantitative data were collected in hard copy format and subsequently entered into the REDCap (Research Electronic Data Capture; Vanderbilt University) system.
Analytical Plan
Basic proportions will be used to assess the rate of nonadherence to each DM complications screening, and 2-screening bundles (eg, either DR+DN, DR+DFC, or DN+DFC screenings) and 3-screening bundles (ie, same-day screenings for all 3 complications; aim 1). Baseline sociodemographic, medical, clinical, and patient-reported factors will be analyzed continuously or categorically where appropriate, and logistic regression models will determine the associations between the various factors with nonadherence to each of the 3 DM complications screening visits, as well as to the 2- and 3- screening bundles (aim 2a). Linear and logistic regression models will be used to examine the relationships between nonadherence to each individual DM complications screening with intermediate pre-post changes in clinical outcomes collected at the follow-up visit (aim 3). To model the economic burden of nonadherence to DM screening, we will estimate and compare the cost of patients who are adherent versus nonadherent (aim 4). We will focus on long-term costs of nonadherence (time horizon: 40 years) as there is unlikely to be any substantial cost-saving in the short term. To estimate long-term costs of nonadherence, a decision tree model will be used, which captures possible benefits of screening adherence in terms of lower risk of developing complications and timely detection and treatment if complications occur which, in turn, would lead to lower costs in the long term. All analyses will be performed using Stata/SE (version 15; StataCorp).
Qualitative Phase
Our qualitative methodology aligns with the Consolidated Criteria for Reporting Qualitative Research (COREQ) checklist (Multimedia Appendix 3) [39].
Population
Focus groups or individual semistructured interviews will be conducted with both adherent and nonadherent participants for the 3 DM complications screenings. Participants will be purposively selected across the spectrum of age, sex, and language spoken (ie, English, Mandarin, Malay, and Tamil) to ensure the applicability of results. In addition, health care professionals, including advanced practice nurses, nurse care managers or educators, and family physicians, will be invited to participate in semistructured interviews.
Sample Size Estimation
Nonadherent participants will be oversampled to focus on barriers to screening adherence as these are amenable to intervention. Interviews will be offered in English, Mandarin, Malay, or Tamil to capture issues applicable to English or non-English speakers. We estimate that 8 focus groups (3 “adherent” and 5 “nonadherent”) with 5 to 7 participants per group, and 15-20 semistructured interviews will be required, resulting in a total of 55-76 participants. We estimate that 8-10 health care professionals will also be needed for the semistructured interviews. Sampling will continue until thematic saturation is reached (ie, no new themes emerge).
Data Collection
Patient focus groups and semistructured interviews will be conducted by trained interviewers and note takers face-to-face at the Singapore Eye Research Institute or over the phone, after the completion of the 16-month follow-up. Sessions will be conducted according to an interview guide developed from a thorough literature review and based on expert panel opinion. The interview guide will comprise open-ended questions targeting predisposing (eg, beliefs about illness and Western and Chinese medicine), reinforcing (eg, role of family members in DM care), and enabling (eg, importance of screening) factors guided by the PRECEDE component (see Multimedia Appendix 4 for the full patient interview guides). Different guides will be used for adherent and nonadherent participants.
We will use the Theoretical Domains Framework (TDF) to inform our interview guide for health care professionals and synthesize our qualitative data (see Multimedia Appendix 4 for full health care professional interview guides) [40]. The TDF covers 12 domains (eg, knowledge, skills, beliefs, and motivation) that cover potentially modifiable barriers that can inform intervention [41]. We will include open-ended questions like “What factors do you think patients consider when they are deciding whether or not to be screened?” as well as questions about knowledge of screening guidelines, referral practices, and information provided to patients regarding the referral during the consultation.
All qualitative interviews will be audio-recorded using a digital recorder. Audio files will be transcribed (and translated into English where necessary) by a professional transcription and translation company and will be checked by the research team for accuracy.
Analytical Plan
An inductive analytical approach will be adopted based on the constant comparative method whereby broad themes will be developed from the raw content of the transcripts [42]. New or improved themes that emerge from later transcripts will be incorporated into the coding hierarchy, and earlier transcripts will be updated to reflect these modifications. Themes will be categorized as predisposing, enabling, and reinforcing barriers or facilitators based on phase 4 of the PRECEDE component. From the health care professionals’ interviews, data relating to factors influencing behavior during complications referral will be coded using the 12 TDF domains. The computer program QSR NVivo 12 will be used to code the data. Transcripts will be analyzed by two different coders and disagreement in coding will be resolved by discussion or a third, independent coder (aim 2b).
Preliminary Analyses
In this preliminary report, we compared the characteristics of participants and those who refused to participate (nonparticipants) using 2-tailed t test or chi-square statistics, as appropriate. Furthermore, we present baseline characteristics that may be associated with nonadherence to DM complications rescreening using descriptive statistics (mean [SD], median [IQR], counts, and percentages). All preliminary analyses were performed using Stata/SE, version 15.
ResultsRecruitment and Retention
Recruitment commenced on June 6, 2018, at Hougang June 11, 2018, at Bedok; June 19, 2018 at Outram; September 25, 2018, at Geylang; November 21, 2018, at Pasir Ris; and February 8, 2019, at Bukit Merah. The baseline quantitative data collection ended at all 6 sites on January 31, 2020.
A total of 3546 patients were screened across the 6 polyclinics. Of these, 830 (23.4%; Figure S2 in Multimedia Appendix 1) were excluded because (1) they received a less than 12-month (approximately 3-6 months) review referral at the polyclinic (n=604, 72.8%); (2) had significant hearing impairment (n=63, 8%); (3) had complications requiring referral to tertiary care (n=39, 5%); (4) had cognitive impairment (n=38, 5%); or (5) were not Chinese, Malay, Indian, or Eurasian (n=35, 4%). Of the remaining 2716 (76.6%) eligible patients, nearly two-thirds refused (n=1742, 64.1%) participation. Reasons for refusal included not being interested (n=901, 51.7%), lack of time (n=512, 29.4%), more than 1 study-related visit required (n=287, 16.5%), and “other” reasons (eg, does not want to share confidential information, the caregiver does not approve, does not believe in research; n=42, 2%). The remaining 974 (35.9%) eligible patients agreed to participate.
The response rate ranged from 9.5% in Bedok and 33.9% in PR to 85.0% in Geylang (Table 1). There were no significant differences in age and sex between participants and nonparticipants. Compared to participants, nonparticipants were more likely to only attend DR screening (n=730, 51.9%), not attend “bundle” screening appointments (n=909, 64.6%) and be from the Bedok site (n=498, 28.6%; all P<.001).
Response rates of eligible patients to the UNADSa study by polyclinic sites.
Polyclinic sites
Eligible patients, n
Participants, n
Response rate, %b
SHPc–Bedok
550
52
9.5
SHP–Outram
638
197
30.9
SHP–Pasir Ris
322
109
33.9
SHP–Bukit Merah
365
57
15.6
NHGPd–Hougang
494
264
53.4
NHGP–Geylang
347
295
85.0
Total
2716
974
35.9
aUNADS: Understanding Non-Adherence to Diabetes Complications Screening.
bCalculated as participants as a percentage of eligible patients.
cSHP: SingHealth Polyclinics.
dNHGP: National Healthcare Group Polyclinics.
Baseline Profile of Study Participants
Of the 974 enrolled patients at baseline, 75.4% (n=734), 61.9% (n=603), and 80.7% (n=786) attended DR, DN, and DFC screening, respectively. Over 4 in 5 (n=793, 81.4%) attended 1 or more complication screenings on the day of enrollment.
Patient characteristics potentially associated with nonadherence to DM complications screening are outlined in the following sections.
Sociodemographics
The baseline cohort had a median age of 61.0 (IQR 55.0-67.0) years, with over two-fifths less than 60 years (n=413, 42.4%); and were primarily male (n=515, 52.9%; Table 2). Most participants were Chinese (n=624, 64.1%), followed by Malay (n=208, 21.4%), Indian (n=136, 13.9%), and Eurasian (n=6, 1%). A majority had received secondary or lower education (n=701, 71.9%), were employed (n=589, 60.5%), had a monthly household income of more than S $2000 (n=518, 62.5%), and lived in public housing (n=843, 86.6%). Furthermore, most were married (n=737, 75.7%) and reported taking less than 30 minutes to access the polyclinic site (n=720, 74.3%).
Baseline sociodemographic and lifestyle characteristics of UNADSa study participants.
Characteristics
Values
Screening attended, n (%)
Diabetic retinopathy
734 (75.4)
Diabetic nephropathy
603 (61.9)
Diabetic foot complication
786 (80.7)
Bundle screening on the same day, n (%)
793 (81.4)
Sociodemographic factors
Age (years), median (IQR)
61.0 (55.0-67.0)
Age groups (years), n (%)
≤49
118 (12.11)
50-59
295 (30.29)
60-69
420 (43.12)
>70
141 (14.48)
Sex, (male), n (%)
515 (52.9)
Ethnicity, n (%)
Chinese
624 (64.1)
Malay
208 (21.4)
Indian
136 (13.9)
Eurasian
6 (0.6)
Educational attainment, n (%)
Primary or lower
153 (15.7)
Secondary or A-level
548 (56.3)
Tertiary
273 (28)
Marital status, n (%)
Single
117 (12)
Married
737 (75.7)
Divorced or separated
61 (6.3)
Widowed
59 (6.1)
Housing type
Public
843 (86.6)
Private
131 (13.5)
Monthly household income
≤S $2000 (US $1524.80)
311 (37.5)
>S $2000
518 (62.5)
Employment status (yes), n (%)
589 (60.5)
Accessibility to polyclinic
Less than 30 minutes
720 (74.3)
30 minutes to 2 hours
249 (25.7)
Lifestyle or health behavior factors
Smoking
Never or past smoker
870 (89.3)
Current smoker
104 (10.7)
Alcohol status
Never or past drinker
763 (78.3)
Current drinker
211 (21.7)
aUNADS: Understanding Non-Adherence to Diabetes Complications Screening.
Medical and Clinical
Comorbidities were common overall, with a high prevalence of hypertension (n=711, 73.4%), dyslipidemia (n=828, 85.1%), and obesity (n=419, 43.1%; Table 3). Most participants rated their health status as “good to excellent” (n=775, 79.7%). Participants reported a median DM duration of 6.3 (IQR 3.0-12.0) years, and most were on diet control (n=915, 94.2%). Participants had median HbA1c, and systolic and diastolic blood pressure values of 6.9% (IQR 6.4%-7.6%), and 130.0 (IQR 120.0-138.0) mm Hg and 70.0 (IQR 65.0-78.0) mm Hg, respectively. The median low-density lipoprotein cholesterol and eGFR was 2.1 (IQR 1.7-2.6) mmol/L and 88.5 (IQR 73.8-98.3) mL/min/1.73m2, respectively.
Baseline medical and clinical characteristics of UNADSa study participants.
Characteristics
Values
Diabetes Management
Diabetes duration, median (IQR)
6.3 (3.0-12.0)
Diabetes treatment, n (%)
Insulin
68 (7)
Oral antidiabetic medications
789 (81.3)
Diet
915 (94.2)
Exercise
780 (80.3)
Health status
Comorbidities
None, n (%)
54 (5.6)
1-2 conditions, n (%)
640 (66.5)
3 conditions or more, n (%)
268 (27.9)
Hypertension (yes), n (%)
711 (73.4)
Dyslipidemia (yes), n (%)
828 (85.1)
Asthma (yes), n (%)
51 (5.2)
BMI (kg/m2)b, median (IQR)
26.9 (24.1-30.3)
Obesity (yes), n (%)
419 (43.1)
Arthritis (yes), n (%)
96 (9.9)
Self-rated health status, n (%)
Excellent or very good
281 (28.9)
Good
494 (50.8)
Fair or poor
198 (20.4)
Clinical factors
Blood pressure (mm Hg), median (IQR)
Systolic
130.0 (120.0-138.0)
Diastolic
70.0 (65.0-78.0)
HbA1cc (%), median (IQR)
6.9 (6.4-7.6)
Lipids (mmol/L), median (IQR)
Total cholesterol
4.1 (3.6-4.7)
HDLd
1.3 (1.1-1.5)
LDLe
2.1 (1.7-2.6)
TGLf
1.4 (1.0-1.8)
eGFRg (ml/min/1.72m2), median (IQR)
88.5 (73.8-98.3)
aUNADS: Understanding Non-Adherence to Diabetes Complications Screening.
bBMI was calculated as weight in kilograms divided by height in meters squared.
cHbA1c: hemoglobin A1c.
dHDL: high-density lipoprotein.
eLDL: low-density lipoprotein.
fTGL: triglyceride level.
geGFR: estimated glomerular filtration rate.
Patient-Reported
In general, most study participants (n=764, 95.0%) had adequate health literacy (Table 4). More than half (n=532, 55.1%) reported not having received DM education either at the time of diagnosis of DM or in the past year. Most reported receiving “quite a lot or a great deal” of DM support (n=678, 69.7%). The median baseline diabetes knowledge scores of those on insulin and not on insulin were 16.0 (IQR 13.0-18.0) out of 22 and 9.0 (IQR 8.0-11.0) out of 14, respectively. Over 10% and 20% of the participants on insulin and not on insulin answered less than 50% of the questionnaire correctly, respectively. Participants reported low levels of diabetes-related self-efficacy (median 1.4, IQR 1.0-3.9 out of 5), and more than half rated their diabetes control status as “okay” (n=547, 56.6%). Almost one-fifth reported diabetes related-distress (n=171, 17.6%).
Baseline patient-reported characteristics of UNADSa study participants.
Characteristics
Values
Health literacyb, n (%)
Not adequate
41 (5.1)
Adequate
764 (95.0)
Health literacy (3-item)c, median (IQR)
10.0 (7.0-12.0)
Diabetes knowledged, median (IQR)
9.0 (8.0-11.0)
Not on insulin treatment
9.0 (8.0-11.0)
On insulin treatment
16.0 (13.0-18.0)
Diabetes educatione (no), n (%)
532 (55.1)
Diabetes support (yes), n (%)
No or a little bit of support
295 (30.3)
Quite a lot or a great deal of support
678 (69.7)
Satisfaction with diabetes support, n (%)
Not at all or a little satisfied
143 (14.7)
Somewhat satisfied
180 (18.5)
Very satisfied or extremely satisfied
648 (66.7)
Self-rated diabetes control, n (%)
Very good
255 (26.4)
Okay
547 (56.6)
Not good
80 (8.3)
Variable
84 (8.7)
Attitude toward glycemic control, n (%)
Very important
916 (94.2)
A little or not important
56 (5.8)
Self-efficacyf, median (IQR)
1.4 (1.0-3.9)
Diabetes-related distressg (yes), n (%)
171 (17.6)
aUNADS: Understanding Non-Adherence to Diabetes Complications Screening.
bAdequate health literacy was defined as 75% correct response in both sections, that is, 3 numeracy items and 27 reading comprehension sections.
cHigher score indicates higher health literacy level.
dHigher score indicates higher level of diabetes knowledge.
eHave you ever attended any diabetes education programs focusing on diabetic retinopathy, diabetic nephropathy, or diabetic foot complications management?
gA total score of 8 or more suggests “possible” diabetes distress.
Discussion
We established the UNADS study to provide urgently needed data on the rates, predictors, and clinical and economic impact of nonadherence to DM complications screening in a multiethnic Asian primary care population with T2DM using a mixed methods approach. A total of 974 patients with T2DM who completed 2123 DM complications screening visits across 6 polyclinics in Singapore were enrolled and have completed the baseline assessment. Most (n=793, 81.4%) were recruited while attending more than 1 complication screening on the same day. A spectrum of sociodemographic (eg, age, ethnicity, education and income levels, housing, and accessibility to nearest polyclinic), medical (eg, presence of comorbidities, and HbA1c and lipid levels), and patient-reported factors (eg, DM-specific self-efficacy and knowledge levels, and self-rated DM control) which may potentially predict nonadherence to DM complications screening, were systematically collected. Completion of longitudinal quantitative and cross-sectional qualitative data collection and analyses will provide insights into the rates and modifiable risk factors of nonadherence to DM complications screening in this population; and contribute to the design of evidence-based interventions addressing those factors at patient, provider, and system levels in the public sector.
We found that more than two-fifths of our study participants were aged less than 60 years, were mostly Chinese, had secondary or lower education, and received a monthly household income of more than S $2000 (US $1546.05). These characteristics have been reported to be linked to nonadherence to DM complications screening by previous studies exploring factors related to attendance to DM-related screening programs [8,43]. For example, a large cohort study (n=24,832) of individuals with DM by Alice et al [8] reported that those aged less than 60 years and receiving a higher income were associated with the greatest risk (–1.3-fold) of nonadherence to eye examination guidelines in New South Wales. Furthermore, a mixed methods study investigating the barriers or facilitators of uptake of DM health screening conducted in Singapore showed that those of Chinese ethnicity (vs Indians), and those with lower education had lower odds of attending regular DM health screenings [43].
Our baseline analysis also showed a high prevalence of comorbid conditions such as hypertension, dyslipidemia, and obesity, which may be associated with an increased likelihood of not attending screening appointments. Comorbidity is common among people with DM and can create a number of challenges that affect a patient’s ability to participate in regular screening [44]. Indeed, Baumeister et al [45] studied trends of barriers to receiving recommended eye care among individuals with DM in Germany using population-based data collected between 1997-2001 and 2008-2012 and found that eye care use declined over the 10-year period in those diagnosed with DM for more than 5 years, and in those with obesity, dyslipidemia, and multiple comorbid conditions.
We found that more than half of our baseline study sample had not received any DM-related education prior to DM complication screening. This is important, as evidence suggests that a lack of information on DM may predict nonadherence to regular annual screening. For instance, in a nationwide survey of 1288 individuals with DM conducted in Korea, Byun et al [12] found that those who had not received education about DM care were more likely to be screened less often for DR and DN. At baseline, our participants reported low levels of DM-specific self-efficacy, which may be important in predicting nonadherence to DM complications screening. Indeed, studies exploring screening attendance in other health care fields (eg, cancer) have shown that low self-efficacy is an important determinant of nonadherence to screening programs [46]. Nevertheless, full analysis of the follow-up data using regression models is essential to identify and confirm the modifiable predictors of DM complications screening in this population and to inform appropriate intervention in this multiethnic primary care population.
A key strength of the UNADS study is the application of the PRECEDE-PROCEED model, which offers a systematic classification of factors at individual, provider, and system levels by their relative importance and capacity for modification. This facilitates consideration of the determinants for change at individual-, provider-, and system-levels and allows for the development of more efficient and effective programs. Other strengths are its large sample size involving six polyclinics across 2 different health care clusters and the collection of rich longitudinal quantitative and qualitative data covering a wide range of exposures and outcomes.
The UNADS study has some important limitations. First, the COVID-19 outbreak and related restrictions had serious implications on its implementation. Due to the restrictions on face-to-face research activities, study enrollment was severely impacted, and the estimated target sample size was not achieved. However, as far more of our study participants attended bundle screening (81.4%) than originally estimated in our sample size calculations (30%), our final sample size was deemed adequate to accomplish our study’s objectives. Secondly, our overall baseline response rate was low (35%) and this could lead to selection bias affecting the validity of our results. There were several reasons for this: survey fatigue experienced by patients attending busy sites like Bedok and Bukit Merah with several ongoing research projects; the need for biological sample collection at baseline and follow-up (if unavailable in the EMR); and the impact of restrictions imposed during the COVID-19 outbreak as explained above [47,48]. Third, we are unable to compare the predictors of nonadherence between 2 different complication screenings (eg, DR vs DFC, DFC vs DN) due to the small number of patients attending both DR and DFC (n=223), and DFC and DN (n=123) screenings. Finally, the use of self-reported measures may be subject to social desirability and recall bias.
The UNADS study is the first longitudinal mixed methods study to investigate the rates, independent predictors, and the clinical and economic impact of nonadherence to DM complications screening in multiethnic polyclinic patients with T2DM in Singapore. Our baseline data are comprehensive including several sociodemographic, clinical, health, and patient-reported characteristics that may predict nonadherence to DM complications screening. Completion of the follow-up phase will provide empirical evidence on the rates, and modifiable barriers or enablers of nonadherence to DM complications screening. Furthermore, the study results may inform public health interventions to develop culturally appropriate intervention programs to improve screening adherence in multiethnic populations. Finally, this study may contribute to reducing unnecessary disease burden with timely screening, monitoring, and treatment; improving health outcomes for Singaporeans with T2DM at the primary care level; and demonstrating significant cost-savings to the patient and society over time.
Supplementary Figures.
Details of the UNADS study procedures, participants and timeline for administration.
Consolidated Criteria for Reporting Qualitative Research (COREQ) checklist.
Additional materials.
The authors thank the UNADS study team for their important contribution to the study implementation: Aminah Hajjah Nawi, Roshinah Seman, Valerie Yeo Jing Jing, Nur Musfirah Binte Mustafah, Lee Jin, Ng Chiat Eng, Ester Lee Pei Xuan, and Melisa. This work was supported by the National Medical Research Council Senior Clinician Scientist Award (NMRC/CIRG/1453/2016) received by ELL (principal investigator). The granting body had no roles in the design, conduct, or data analysis of the study, nor in manuscript preparation and approval. Generative artificial intelligence was not used in this study.
Data Availability
The datasets generated and analyzed as part of this project will be available from the corresponding author on reasonable request after the research team has completed analyses related to the study aims.
AA was responsible for conducting the formal analysis and investigation, managing the project, supervising the research process, drafting the original manuscript, and contributing to its review and editing. EF contributed to the conceptualization of the study, carried out formal analysis and investigation, managed project administration, supervised the research, and was involved in drafting, reviewing, and editing the manuscript. AW participated in formal analysis. EL, RM, and CS contributed to the conceptualization, and review and editing of the manuscript. NC, WE, CJ, LP, WT, WF, and SW were involved in the writing process, specifically in reviewing and editing the manuscript. EL served as the senior study investigator and lead author. The final version of the paper has been seen and approved by all coauthors.
None declared.
Abbreviations
BP: blood pressure
COREQ: Consolidated Criteria for Reporting Qualitative Research
DFC: diabetic foot complication
DM: diabetes mellitus
DN: diabetic nephropathy
DR: diabetic retinopathy
eGFR: estimated glomerular filtration rate
EMR: electronic medical record
HbA1c: hemoglobin A1c
NHGP: National Healthcare Group Polyclinics.
OR: odds ratio
PRECEDE: Predisposing, Reinforcing, and Enabling Constructs in Educational Diagnosis and Evaluation
PROCEED: Policy, Regulatory, and Organizational Constructs in Educational and Environmental Development
REDCap: Research Electronic Data Capture
SHP: SingHealth Polyclinics
T2DM: type 2 diabetes mellitus
TDF: Theoretical Domains Framework
UNADS: Understanding Non-Adherence to Diabetes Complications Screening.
HardingJLPavkovMEMaglianoDJShawJEGreggEWGlobal trends in diabetes complications: a review of current evidenceDiabetologia201962131610.1007/s00125-018-4711-23017127910.1007/s00125-018-4711-2TanKWDickensBSLCookARProjected burden of type 2 diabetes mellitus-related complications in Singapore until 2050: a Bayesian evidence synthesisBMJ Open Diabetes Res Care202081e00092810.1136/bmjdrc-2019-000928321842038/1/e000928PMC7076230AgardhEAgardhCHansson-LundbladCThe five-year incidence of blindness after introducing a screening programme for early detection of treatable diabetic retinopathyDiabet Med199310655555910.1111/j.1464-5491.1993.tb00120.x8365093AngGYYapCWSaxenaNEffectiveness of diabetes foot screening in primary care in preventing lower extremity amputationsAnn Acad Med Singap2017461141742310.47102/annals-acadmedsg.v46n11p417DelevryDHoALeQAAssociation between processes of diabetes care and health care utilization in patients with diabetes: evidence from a nationally representative US sampleJ Diabetes2021131788810.1111/1753-0407.1310932851797ImaiCLiLHardieRGeorgiouAAdherence to guideline-recommended HbA1c testing frequency and better outcomes in patients with type 2 diabetes: a 5-year retrospective cohort study in Australian general practiceBMJ Qual Saf202130970671410.1136/bmjqs-2020-01202633542066bmjqs-2020-012026PMC8380884PalmerAJValentineWJChenRMehinNGabrielSBregmanBRodbyRAA health economic analysis of screening and optimal treatment of nephropathy in patients with type 2 diabetes and hypertension in the USANephrol Dial Transplant20082341216122310.1093/ndt/gfn08218359872gfn082GibsonAAHumphriesJGilliesMNassarNColagiuriSAdherence to eye examination guidelines among individuals with diabetes: an analysis of linked health dataClin Exp Ophthalmol20204891229123810.1111/ceo.1383032710452RimTHTByunIHKimHSLeeSYYoonJSFactors associated with diabetic retinopathy and nephropathy screening in Korea: the Third and Fourth Korea National Health and Nutrition Examination Survey (KNHANES III and IV)J Korean Med Sci201328681482010.3346/jkms.2013.28.6.81423772143PMC3677995TappRJZimmetPZHarperCAde CourtenMPBalkauBMcCartyDJTaylorHRWelbornTAShawJEDiabetes care in an Australian population: frequency of screening examinations for eye and foot complications of diabetesDiabetes Care200427368869310.2337/diacare.27.3.68814988286HuangOSTayWTOngPGSabanayagamCChengCTanGSCheungGCMLamoureuxELWongTYPrevalence and determinants of undiagnosed diabetic retinopathy and vision-threatening retinopathy in a multiethnic Asian cohort: the Singapore Epidemiology of Eye Diseases (SEED) studyBr J Ophthalmol201599121614162110.1136/bjophthalmol-2014-30649225953847bjophthalmol-2014-306492ByunSMaSHJunJKJungKParkBScreening for diabetic retinopathy and nephropathy in patients with diabetes: a nationwide survey in KoreaPLoS One201385e6299110.1371/journal.pone.006299123667557PONE-D-13-01878PMC3648467EppleySEMansbergerSLRamanathanSLowryEACharacteristics associated with adherence to annual dilated eye examinations among US patients with diagnosed diabetesOphthalmology2019126111492149910.1016/j.ophtha.2019.05.03331281055S0161-6420(19)30130-7FayfmanMSchechterMCAmobiCNWilliamsRNHillmanJLAlamMMRajaniRRZiemerDCKempkerRRUmpierrezGEBarriers to diabetic foot care in a disadvantaged population: a qualitative assessmentJ Diabetes Complications2020341210768810.1016/j.jdiacomp.2020.10768832917487S1056-8727(20)30450-5PMC10424493HipwellAESturtJLindenmeyerAStrattonIGadsbyRO'HarePScanlonPHAttitudes, access and anguish: a qualitative interview study of staff and patients' experiences of diabetic retinopathy screeningBMJ Open2014412e00549810.1136/bmjopen-2014-00549825510885bmjopen-2014-005498PMC4267079HungSLLFuSNLauPSWongSYSA qualitative study on why did the poorly-educated Chinese elderly fail to attend nurse-led case manager clinic and how to facilitate their attendanceInt J Equity Health2015141010.1186/s12939-015-0137-325636857s12939-015-0137-3PMC4322441van EijkKNDBlomJWGusseklooJPolakBCPGroeneveldYDiabetic retinopathy screening in patients with diabetes mellitus in primary care: incentives and barriers to screening attendanceDiabetes Res Clin Pract2012961101610.1016/j.diabres.2011.11.00322137363S0168-8227(11)00626-7SachdevaAStrattonIUnwinJMoretonRScanlonPDiabetic retinopathy screening: Study to determine risk factors for non-attendanceDiabetes Prim Care2012145308316Garreta RufasAMeredithKHarrisJRossingPHobbsRWannerCMO369: A systematic review to evaluate the drivers of non-adherence to albuminuria testing guidelines and the clinical and economic impact of not identifying CKD over the course of progressive kidney function lossNephrology Dialysis Transplantation20223710.1093/ndt/gfac069.002KashimRNewtonPOjoODiabetic retinopathy screening: A systematic review on patients' non-attendanceInt J Environ Res Public Health201815115710.3390/ijerph1501015729351207ijerph15010157PMC5800256LeeseGPBoylePFengZEmslie-SmithAEllisJDScreening uptake in a well-established diabetic retinopathy screening program: the role of geographical access and deprivationDiabetes Care200831112131213510.2337/dc08-109818728235dc08-1098PMC2571062GreenLKreuterMHealth Program Planning: An Educational and Ecological Approach2005New York, NYMcGraw-HillSaulleRSinopoliADe Paula BaerAMannocciAMarinoMDe BelvisAGFedericiALa TorreGThe PRECEDE-PROCEED model as a tool in public health screening: a systematic reviewClin Ter20201712e167e17710.7417/CT.2020.220832141490StudtsCRTarasenkoYNSchoenbergNEBarriers to cervical cancer screening among middle-aged and older rural appalachian womenJ Community Health201338350051210.1007/s10900-012-9639-823179390PMC3600402TrammRMcCarthyAYatesPUsing the precede-proceed model of health program planning in breast cancer nursing researchJ Adv Nurs20126881870188010.1111/j.1365-2648.2011.05888.x22142482Unger-SaldañaKSaldaña-TellezMPotterMBVan LoonKAllen-LeighBLajousMBarriers and facilitators for colorectal cancer screening in a low-income urban community in Mexico CityImplement Sci Commun202016410.1186/s43058-020-00055-z3288521955PMC7427948TohMLeeLThamLQuality care of diabetes mellitus in the Singapore national healthcare group: 5-year trend from 2008 to 20122014The 10th IDF-WPR Congress and 6th AASD Scientific MeetingDecember 4, 2014Ulaanbaatar, Mongolia10.1016/s0168-8227(14)70407-3BrookePBullockRValidation of a 6 item cognitive impairment test with a view to primary care usageInt J Geriat Psychiatry1999141193694010.1002/(sici)1099-1166(199911)14:11<936::aid-gps39>3.0.co;2-1ManolioTAFishelSCBeattieCTorresJChristophersonRMerrittWTWheltonPKEvaluation of the dinamap continuous blood pressure monitorAm J Hypertens19881161S167S10.1093/ajh/1.3.161s3415794LeveyASStevensLASchmidCHZhangYCastroAFFeldmanHIKusekJWEggersPVan LenteFGreeneTCoreshJA new equation to estimate glomerular filtration rateAnn Intern Med2009150960461210.7326/0003-4819-150-9-200905050-0000619414839150/9/604PMC2763564FitzgeraldJTFunnellMMAndersonRMNwankwoRStansfieldRBPiattGAValidation of the revised brief Diabetes Knowledge Test (DKT2)Diabetes Educ201642217818710.1177/0145721715624968267697570145721715624968ZainudinSAngDSohAKnowledge of diabetes mellitus and safe practices during Ramadan fasting among Muslim patients with diabetes mellitus in SingaporeSingapore Med J201758524625210.11622/smedj.201608527121922j58/5/246PMC5435842KoYLeeJYTohMPHSTangWTanASDevelopment and validation of a general health literacy test in SingaporeHealth Promot Int2012271455110.1093/heapro/dar02021467097dar020ChewLDBradleyKABoykoEJBrief questions to identify patients with inadequate health literacyFam Med20043685889415343421AndersonRMFitzgeraldJTGruppenLDFunnellMMOhMSThe Diabetes Empowerment Scale-Short Form (DES-SF)Diabetes Care20032651641164210.2337/diacare.26.5.1641-a12716841McGuireBEMorrisonTGHermannsNSkovlundSEldrupEGagliardinoJKokoszkaAMatthewsDPibernik-OkanovićMRodríguez-SaldañaJde WitMSnoekFJShort-form measures of diabetes-related emotional distress: the Problem Areas in Diabetes Scale (PAID)-5 and PAID-1Diabetologia2010531666910.1007/s00125-009-1559-519841892SpeightJBrowneJLHolmes-TruscottEHendrieckxCPouwerFDiabetes MILES—Australia (management and impact for long-term empowerment and success): methods and sample characteristics of a national survey of the psychological aspects of living with type 1 or type 2 diabetes in Australian adultsBMC Public Health20121212010.1186/1471-2458-12-120223250321471-2458-12-120PMC3312855TangTSBrownMBFunnellMMAndersonRMSocial support, quality of life, and self-care behaviors among African Americans with type 2 diabetesDiabetes Educ200834226627610.1177/01457217083156801837577634/2/266TongASainsburyPCraigJConsolidated Criteria for Reporting Qualitative Research (COREQ): A 32-item checklist for interviews and focus groupsInt J Qual Health Care200719634935710.1093/intqhc/mzm04217872937mzm042MichieSJohnstonMAbrahamCLawtonRParkerDWalkerA"Psychological Theory" GroupMaking psychological theory useful for implementing evidence based practice: a consensus approachQual Saf Health Care2005141263310.1136/qshc.2004.0111551569200014/1/26PMC1743963FrenchSDGreenSEO'ConnorDAMcKenzieJEFrancisJJMichieSBuchbinderRSchattnerPSpikeNGrimshawJMDeveloping theory-informed behaviour change interventions to implement evidence into practice: a systematic approach using the theoretical domains frameworkImplement Sci201273810.1186/1748-5908-7-38225310131748-5908-7-38PMC3443064RicePEzzyDQualitative Research Methods: A Health Focus1999Oxford, UKOxford University PressAshaRaniPVDeviFWangPAbdinEZhangYRoystonnKJeyagurunathanASubramaniamMFactors influencing uptake of diabetes health screening: a mixed methods study in Asian populationBMC Public Health2022221151110.1186/s12889-022-13914-23594157910.1186/s12889-022-13914-2PMC9360713PietteJDKerrEAThe impact of comorbid chronic conditions on diabetes careDiabetes Care200629372573110.2337/diacare.29.03.06.dc05-20781650554029/3/725BaumeisterSESchomerusGAndersenRMTostFMarkusMRPVölzkeHJürgensCTrends of barriers to eye care among adults with diagnosed diabetes in Germany, 1997-2012Nutr Metab Cardiovasc Dis2015251090691510.1016/j.numecd.2015.07.00326298427S0939-4753(15)00176-3BongaertsTHBüchnerFLMiddelkoopBJGuicheritORNumansMEDeterminants of (non-)attendance at the Dutch cancer screening programmes: a systematic reviewJ Med Screen202027312112910.1177/096914131988799631801039PMC7491249BixoLCunninghamJLEkseliusLÖsterCRamklintM'Sick and tired': Patients reported reasons for not participating in clinical psychiatric researchHealth Expect202124 Suppl 1Suppl 1202910.1111/hex.1297731605443PMC8137497VillarosaARRamjanLMManezeDGeorgeAConducting population health research during the COVID-19 pandemic: impacts and recommendationsSustainability2021136332010.3390/su13063320