Article Text

Download PDFPDF

Original research
Evaluation of the clinical effectiveness of telephone consultation compared to face-to-face consultation in terms of glycaemic control among patients with suboptimally controlled type 2 diabetes: a retrospective cohort study
  1. Zhong Wei Jeremy Koh1,
  2. Sai Zhen Sim2,
  3. Kaiwei Jeremy Lew2,
  4. Poay Sian Sabrina Lee2,
  5. Eng Sing Lee2
  1. 1National Healthcare Group Polyclinics, Singapore
  2. 2Clinical Research Unit, National Healthcare Group Polyclinics, Singapore
  1. Correspondence to Dr Zhong Wei Jeremy Koh; jeremy_zw_koh{at}nhgp.com.sg

Abstract

Objective With the COVID-19 pandemic, telemedicine has been increasingly deployed in lieu of face-to-face consultations for management of diabetes in primary care. There was a need to evaluate clinical effectiveness of telephone consultations for diabetes management and this study aimed to show whether one-off telephone consultation was inferior or not to face-to-face consultation in terms of glycaemic control among patients with suboptimally controlled type 2 diabetes.

Design Retrospective cohort study. Data of all patients with type 2 diabetes who had a chronic disease consultation during the period 9 April 2020–18 September 2020, and met the study’s inclusion and exclusion criteria was obtained from the electronic medical records.

Setting A primary care clinic in the north-eastern region of Singapore. The clinic’s patient population was representative of Singapore’s population in terms of gender and age.

Participants 644 patients with type 2 diabetes and glycated haemoglobin (HbA1c) 7.0% and above, aged 21–80 years old.

Interventions Participants either underwent telephone or face-to-face consultation for diabetes management.

Outcome measure Mean HbA1c change (∆HbA1c) between preintervention and postintervention.

Results Over 4 months, the mean ∆HbA1c was −0.16 percentage points (p.p.) (95% CI −0.26 to –0.07) and −0.11 p.p. (95% CI −0.20 to –0.02) for face-to-face and telephone consultation groups, respectively. The difference in mean ∆HbA1c between the two groups was +0.05 p.p. (95% CI −∞ to 0.16), with the upper limit of the one-sided 95% CI less than the prespecified non-inferiority margin of 0.5 p.p. (p<0.05). In those with HbA1c≥9%, the difference in mean ∆HbA1c was +0.31 p.p. (95% CI −∞ to 0.79), which exceeded the non-inferiority margin.

Conclusion For patients with suboptimally controlled type 2 diabetes, one-time telephone consultation was non-inferior to face-to-face consultation in terms of glycaemic control in the short term. However, more studies are required to investigate the long-term effects of telephone consultations and for those with HbA1c≥9%.

  • telemedicine
  • primary care
  • general diabetes

Data availability statement

Data are available on reasonable request. Data are available on reasonable request. Data are not available for online access. Readers who wish to gain access to the data can write to the senior author ESL at NHGP_CRU@nhgp.com.sg with their requests. Access can be granted subject to approval of the National Healthcare Group Domain Specific Review Board (DSRB) and in line with the National Healthcare Group Research Data Policy. This is a requirement mandated for this research study by our DSRB and funders.

http://creativecommons.org/licenses/by-nc/4.0/

This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.

Statistics from Altmetric.com

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

STRENGTHS AND LIMITATIONS OF THIS STUDY

  • This study performed propensity score matching to control for known confounding factors.

  • The clinical data from the electronic medical records reflected real-world clinical conditions.

  • A limitation was unobserved confounders may not have been addressed.

  • The study was performed in one primary care clinic and the sample population may not be completely representative of all the patients with diabetes in Singapore.

Introduction

Telemedicine refers to the systematic provision of healthcare services over physically separate environments via information and communications technology.1 There are several modes of telemedicine, including mobile phone text messaging, consultations via videoconferencing and t internet-based glucose monitoring programmes. In Singapore, prior to the pandemic, telemedicine experience in primary care was mostly limited to management of simple acute conditions over video systems by private general practitioners. Telemedicine has not been used for managing chronic diseases due to the close proximity of a primary care clinic to most of the population. On the other hand, telemedicine has been studied for chronic disease management in other countries for several years.2

Singapore confirmed its first case of COVID-19 on 23 January 2020. To prevent the spread of infections, social distancing measures were implemented throughout the country. Subsequently, due to the surge in infection cases, Singapore went into a lockdown in April 2020. This had an impact on the medical care of people with chronic diseases, including those with diabetes. Many of these patients were concerned about attending appointments in the clinic, especially as diabetes was a risk factor for higher morbidity and mortality from COVID-19.3 In Singapore, the prevalence of diabetes was 6.9% in 2019,4 and it was the second highest risk factor for death and disability combined.5 Hence, it was imperative to continue to provide adequate care for patients with diabetes and prevent complications. There was a need to cater for patients with chronic diseases, as primary care clinics sought ways to reduce physical face-to-face consultations, allow patients to receive care at home and thereby prevent crowding in clinics. As a result, primary care turned to telemedicine in place of face-to-face consultations for chronic disease management.6

There have been many studies evaluating the effectiveness of telemedicine in diabetes care. However, as there are so many forms of telemedicine, the results of previous studies have been varied.7–11 One of the advantages of telemedicine is that it allows more frequent monitoring and follow-up assessments by the healthcare team, and thus it can improve glycaemic control.2 12–15 Many studies on telemedicine in diabetes management had incorporated telemonitoring, which required patients to collect and upload their own diabetes-related data on an electronic platform. A member of the diabetes care team would then review the data and provide appropriate medical advice.6 Other studies using various technological portals such as email, videoconference and internet web pages were also found to improve glycaemic control.6 16 These interventions are, however, quite different from the physician-led telephone consultations carried out for routine diabetes care in our setting.17 18

With the pandemic on us, we initiated telephone consultations for patients with diabetes. These consultations were in place of regular face-to-face clinic consultations and did not entail the home glucose monitoring as described above. Hence, there was a need to evaluate its effectiveness compared with face-to-face consultations. In this study, we aimed to evaluate the clinical effectiveness of telephone compared with face-to-face consultation regarding the clinical parameter of glycaemic control, among patients with suboptimally controlled type 2 diabetes. The objective is to analyse the mean changes in glycated haemoglobin (HbA1c) between the two groups.

Methods

Study design

We conducted a retrospective propensity score matched cohort study at Hougang Polyclinic, using data from 9 April 2020 to 30 November 2020. Propensity score matching was used to reduce bias due to confounding variables. Polyclinics are one-stop primary care centres providing a range of subsidised health services, from treatment of acute conditions to management of chronic diseases, and women and child health services. Hougang Polyclinic is part of the National Healthcare Group Polyclinics (NHGP), serving close to 227 000 residents in the north-eastern region of Singapore. Compared with the population of Singapore, this region’s residents are similar in terms of gender and age, but there is a greater proportion of ethnic Chinese and smaller proportion of ethnic Malays living in this region. Anonymised data were retrieved from the NHGP electronic medical records.

Patient and public involvement

Patients or the public were not directly involved in the design, conduct or the reporting or dissemination plans of this study.

Clinic intervention (telephone consultation)

Due to the COVID-19 pandemic, social distancing measures were in place and telephone consults were conducted instead of regular face-to-face consults for eligible patients. Patients who had clinical laboratory tests done weeks or days prior to their consultation date were offered telephone consultation and recruited if they were agreeable, had no recent hospitalisation in the last 6 months, and no hearing impairment. Patients who had face-to-face consultation were those who rejected the telephone consultation or who had clinical laboratory tests scheduled on the same day as their face-to-face consultation.

Patient data selection

We included all patients with type 2 diabetes who had a consultation for chronic disease management during the period 9 April 2020–19 September 2020, and then applied the inclusion and exclusion criteria as listed below to obtain the study population The first consultation that the patient had during the study period is defined as the index consultation. This index consultation could either be a face-to-face or telephone consultation. The inclusion criteria included patients with type 2 diabetes, age between 21 and 80 years old, baseline HbA1c≥7.0% at index consultation, and follow-up HbA1c 3–4 months after the index consultation. We excluded patients if they returned to the clinic for any other appointments during the study period, had no baseline HbA1c within the 2 weeks prior to the index consultation, had follow-up HbA1c earlier than 3 months or later than 4 months from the index consultation or had missing clinical indicators. As the life span of red blood cells in the human body is around 120 days, the HbA1c provides an indication of blood glucose control over the same period.19 20 Hence a follow-up duration of up to 4 months was selected.

Data collection for propensity score matching

The data collected included demographics such as age, gender and ethnicity and clinical indicators such as body mass index (BMI), blood pressure, low density lipoprotein (LDL), estimated glomerular filtration rate (eGFR), urinary albumin-creatinine ratio and number of diabetic medications. The presence of other chronic conditions was retrieved as well, namely comorbid conditions such as hypertension and hyperlipidaemia, diabetic macrovascular conditions such as ischaemic heart disease, stroke and peripheral vascular disease, and diabetic microvascular conditions such as chronic kidney disease, diabetic retinopathy and neuropathy.

Outcome

The main outcome measure was the mean HbA1c change (∆HbA1c) between the two visits. The first HbA1c test would be performed within 2 weeks prior to the index consultation and subsequent HbA1c test done 3–4 months later.

Statistical analysis

Propensity score analysis was performed to assess differences between the telephone and face-to-face consultation. The patients in the telephone consultation group were matched 1:1 to those in the face-to-face consultation group based on propensity scores. Variables used for matching include age, gender, ethnicity, clinical indicators, the presence of comorbid conditions scuh as hyperlipidaemia, hypertension, macrovascular and microvascular diabetic complications. The subgroups of the clinical indicators were based on clinical targets from existing local guidelines for the particular indicator. For example, BMI of <23 kg/m2 indicates low health risk, between 23 and 27.5 indicates moderate health risk and ≥27.5 indicates high health risk. Blood pressure ≥140/80 mm Hg and LDL of ≥2.6 mmol/L are above guideline target levels as well. A 1-to-1 matching algorithm without replacement was adopted. We evaluated the balances of matched covariates with standardised mean differences and considered differences of less than 0.2 to be matched sufficiently.21 22 Besides the primary outcome, a subgroup analysis stratified according to the baseline HbA1c was also performed.

We used a one-sided Welch’s independent samples t-test to test for non-inferiority.

Non-inferiority is claimed if the test is statistically significant (p<0.05) or if the upper limit of the one-sided 95% CI did not exceed 0.5 percentage points (p.p.) (clinically relevant change in HbA1c).13

Our null (H0) and alternative (H1) hypotheses are stated as follows:

Embedded Image

Embedded Image

where Embedded Image and Embedded Image are the mean ∆HbA1c of the telephone and face-to-face consultation groups, respectively.

Results

There were 14 776 patients with diabetes who had a consultation with regard to type 2 diabetes at the polyclinic between 9 April 2020 and 14 September 2020. We included 1276 patients who fulfilled the selection criteria (figure 1). A total of 954 patients had undergone a face-to-face consultation and 322 had undergone a telephone consultation. Many important baseline factors were similar between the two groups. For face-to-face and telephone consultations, the gender distribution, mean blood pressures and mean LDL levels were 51.7% and 52.8% male, 130/72 mm Hg and 130/73 mm Hg, and 2.22 mmol/L and 2.28 mmol/L, respectively (table 1). For the face-to-face consultation group, 21.6% had at least one diabetic macrovascular complication and 67.9% had at least one diabetic microvascular complication, compared with 20.2% and 65.8%, respectively, in the telephone consultation group.

Table 1

Characteristics of study population before propensity matching

Figure 1

Flow chart for cohort selection. HbA1c, glycated haemoglobin.

However, there were significant differences for mean age (62.85 years vs 60.61 years) and baseline HbA1c (8.46% vs 8.03%) for face-to-face and telephone consultation groups, respectively. Table 1 shows the baseline demographics of the patients in the unmatched cohorts. Before propensity score matching, there was a significant difference between the two groups. The standardised mean difference was greater than 0.2 for mean age, baseline HbA1c, eGFR and diabetic medication count.

After propensity score matching, the groups were well-balanced and the standardised mean difference was less than 0.2 in all covariates. In total, 322 patients in the telephone consult group were matched to 322 patients in the face-to-face consult group (table 2).

Table 2

Characteristics of propensity matched population

Among the face-to-face consultation group, the mean HbA1c change was −0.16 p.p. (95% CI −0.26 to –0.07), compared with the telephone consult group where the mean HbA1c change was −0.11 p.p. (95% CI −0.20 to –0.02). The difference in mean change in HbA1c between the two groups was +0.05 p.p. (95% CI −∞ to 0.16). The upper limit of the one-sided 95% CI was less than the prespecified non-inferiority margin of 0.5 p.p., which met the criterion for declaring non-inferiority of telephone consult to face-to-face consult for glycaemic control among patients with suboptimally controlled type 2 diabetes (figure 2).

Figure 2

Assessment of non-inferiority (non-inferiority margin set at 0.5 p.p.). ∆HbA1c, mean glycated haemoglobin change; p.p., percentage points.

The difference in mean change in HbA1c also varied among the various subgroups of baseline diabetic control. It was +0.31 p.p. (95% CI −∞ to 0.79)) for HbA1c≥9%, +0.03 p.p. (95% CI −∞ to 0.19) for HbA1c 8.0–8.9% and +0.02 p.p. (95% CI −∞ to 0.15) for HbA1c 7.0%–7.9%. For HbA1c≥9%, the upper limit of the one-sided 95% CI was more than the prespecified non-inferiority margin of 0.5 p.p. This indicates telephone consultation was inferior to face-to-face consultation in these patients. For the other two subgroups, the upper limit of the one-sided 95% CI was less than the prespecified non-inferiority margin of 0.5 p.p., indicating non-inferiority of telephone consultation to face-to-face consultation for those with a baseline HbA1c of 7.0%–8.9% (figure 2).

Discussion

This study evaluated the effectiveness of one-time telephone consultation compared with face-to-face consultation for patients with type 2 diabetes who had a baseline HbA1c of 7% and above. In terms of glycaemic control, telephone consultation was found to be non-inferior to face-to-face consultation in this group of patients. In the subgroup analysis, telephone consultation appears to be inferior to face-to-face consultation for patients with a HbA1c of 9% and above.

As far as we know, this is the first study in Singapore to assess telephone consultations for type 2 diabetes. Telephones are ubiquitous and consultations over the telephone are more accessible and readily accepted by the elderly or those less technologically inclined as no special equipment or mobile applications are needed. In a meta-analysis done by Lee SWH, it was found that telemedicine strategies can be useful and lead to clinically meaningful reduction in HbA1c.18 However, telemedicine modalities are very varied and only 8 out of the 107 studies included in the analysis had performed teleconsultation. This study adds to the body of evidence of the utility of telephone consultations in the management of patients with diabetes. Due to the ongoing pandemic, other telemedicine modalities are being initiated as well.23 For videoconsultations, several studies have found that they improved HbA1c results compared with face-to-face consultations over months.24–26 Videoconsultations, however, are inherently different from telephone consultations as there is eye contact between the patient and physician which may build better rapport than a consultation over the phone. However, videoconferencing can be a barrier for the less technologically inclined.

In a meta-analysis performed by Lee et al, greater improvements in glycaemic control with telemedicine were reported in studies where patients had a mean baseline HbA1c level of 8% and greater.10 Our study produced different results where patients with HbA1c of ≥8% at the index consultation and underwent telephone consultations did not have an improvement in glycaemic control. However, our study was not designed to detect superiority, and further studies should be done to investigate this further.

Type 2 diabetes is a chronic disease and long-term good glycaemic control is essential to prevent development of diabetic complications. This study has provided reassurance that in the current COVID-19 pandemic, telephone consultations can be employed as a short-term measure, in line with social distancing measures and prevent overcrowding in clinics. This was also shown in a systematic review done in 2017 by Faruque et al that telemedicine may be a useful supplement to control HbA1c in the short term.11

However, its long-term outcomes are not known. For example, if a patient underwent telephone consultations repeatedly over a year rather than face-to-face consultations, it is unknown what his glycaemic control will be like. Besides glycaemic control, there are other aspects of diabetes care that are important as well including control of comorbid conditions such as hyperlipidaemia and hypertension and screening of complications like retinopathy and neuropathy. Other factors such as patient and physician satisfaction or cost-effectiveness were also not studied.

The main strength of this study was that propensity score matching was performed to control the known confounding factors between the two groups as much as possible. The covariates of the two groups were balanced before analysis was carried out. The clinical data from the electronic medical records also reflected real-world clinical conditions.

However, the study has some limitations. First, due to the retrospective nature of the study, some unobserved confounders may not have been addressed. Propensity score matching was carried out but there may still be some form of selection bias. Second, this study was carried out in one primary care clinic and hence the results of the study may not be generalisable to the rest of the country. Third, the data retrieved is based on what can be recorded on the medical records system. There are certain demographic details that are not present on the system such as education level, occupation and income. Some of the medical conditions may not have been coded accurately as well. Fourth, this study assessed the effect of only one telephone consultation over a short period of 4 months. Hence, the longer-term effects need to be studied further. Fifth, in the subgroup analysis which showed that telephone consultations may be inferior to face-to-face consultations for those with HbA1c≥9%, this result was limited by its sample size (11.5% of the whole cohort) and further studies should be carried out to investigate this further.

Conclusion

Telephone consultations for suboptimally controlled type 2 diabetics are non-inferior to face-to-face consultations. Further studies should be performed to evaluate if this is holds true in the longer term. Our results suggest that telephone consultations may not be as effective for patients with HbA1c of 9.0% and above, this should be further studied to delineate its effectiveness in poorly controlled diabetes. This study forms a starting point for telemedicine for chronic disease management in Singapore, and further studies into other modalities like videoconsultations should be considered.

Data availability statement

Data are available on reasonable request. Data are available on reasonable request. Data are not available for online access. Readers who wish to gain access to the data can write to the senior author ESL at NHGP_CRU@nhgp.com.sg with their requests. Access can be granted subject to approval of the National Healthcare Group Domain Specific Review Board (DSRB) and in line with the National Healthcare Group Research Data Policy. This is a requirement mandated for this research study by our DSRB and funders.

Ethics statements

Patient consent for publication

Ethics approval

Ethical approval was obtained from National Healthcare Group Domain Specific Review Board (NHG DSRB) on 15 December 2020 (reference number 2020/01237).

Acknowledgments

We would like to thank Xie Ying from the Information Management and Analytics Department at National Healthcare Group Polyclinics, who assisted with data collection.

References

Footnotes

  • Contributors ZWJK, SZS, KJL, PSSL and ESL initiated and conceptualised the study; ZWJK and KJL developed the protocol of the study with the help of SZS, PSSL and ESL. KJL performed the statistical analysis. ZWJK wrote the first draft of manuscript with inputs from SZS and ESL. The draft was revised by ZWJK, KJL, SZS, PSSL and ESL. ZWJK acted as guarantor of the study.

  • Funding This research was supported by the Singapore Ministry of Health’s National Medical Research Council under the Centre Grant Programme (reference number: NMRC/CG/C019/2017).

  • Disclaimer The funding agency was not involved in the design of the study, collection, analysis and interpretation of data and writing of the manuscript.

  • Competing interests None declared.

  • Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.

  • Provenance and peer review Not commissioned; externally peer reviewed.