Quality in Primary Care Open Access

Keywords

chronic disease, patient adherence, reminder letters

Introduction

Research has showed that American adult patients receive the appropriate clinical care just over half of the time for a wide variety of recommended services.[1] With so many Americans not receiving these recommended services, there has been much effort to identify the factors contributing to the low adherence rates, as well as an increased interest in finding ways to improve upon current adherence rates.[2] Both patient and healthcare system-level factors have been suggested as important components. For example, patient-level factors include: a lack of patient awareness of the need for the recommended services; embarrassment involved in completing the services; fear of pain or other sideeffects from the recommended services; patients feeling asymptomatic or less susceptible to diseases; and a lack of insurance coverage.[3–9] For the healthcare systemfactors, the Institute of Medicine’s (IOM’s) landmark publication Crossing the Quality Chasm suggests that the American system is overly complex and uncoordinated, requiring multiple steps and patient ‘hand-offs’ that decrease patient care efficiency and compromise patient safety.[10] Patients additionally identify healthcare system barriers as ‘poor access to services and unaffordable costs’.[9]

There are several reported interventions that have aimed to improve patients’ adherence to their recommended care. Current research has focused on three dominant patient reminder strategies: letters, phone calls and text messaging – with the letter reminder strategy being the most commonly studied intervention. The letter reminder strategy has been shown to be cost-effective, widely accepted and welcomed by patients.[11–13] Yet, although there have been many strategies studied to improve adherence rates, most have focused on preventative services and chronic disease management, with significant variability reported. Furthermore, research has not yet focused on symptom specific recommendations, such as referrals for acute complaints (i.e. eye pain or abdominal pain). There is also a paucity of literature delineating any organisational or system-level characteristics or strategies aimed at improving adherence to the recommended care.

This study analysed pilot data from one physician’s office at one academic, urban healthcare organisation that utilised the letter reminder strategy. The goal of this study was to determine whether there were differences between preventative and chronic disease adherence rates and symptom-specific patient adherence rates, as well as to explore the extent to which there was a relationship between the complexity of that healthcare organisation’s processes with adherence rates. We hypothesised that adherence rates for preventative and chronic disease management would be similar to rates found in prior research, and that symptom-specific recommendations would show an improvement from the letter reminder strategy. We further hypothesised that the complexity of the healthcare system processes would generally be inversely related to the adherence rates.

Method

Data were obtained from August 2008 to May 2009 from one primary care physician’s records at an academic, urban healthcare centre. All patients were provided with appropriate counselling for risks and benefits of the ordered test or referrals. All recommended tests or referrals agreed upon by the patient and ordered by the physician were entered into a database and reconciled monthly for completion. The patients of this practice were all adults older than 18 years and had a variety of health insurances including Health Maintenance Organization (HMO), Preferred Provider Organization (PPO) and Medicare.

The physician’s office followed a standard protocol for reconciling the completion of recommended services as well as for mailing reminder letters. Initial reconciliation and mailing of a first reminder letter occurred after one month for urgent recommendations and two months for non-urgent recommendations. A second reminder letter was mailed one month after the first reminder letter if the ordered service was not completed at each monthly reconciliation process following the mailing of the first reminder letter.

Upon reviewing the dataset, ordered services were divided into several categories: (1) Prevention and Chronic Disease Management; (2) Referrals (acute, symptom-specific management); (3) Radiology and Diagnostic Testing (acute, symptom-specific management); and (4) Labs (acute and non-acute). Prevention and Chronic Disease Management included tests for colon cancer screening (colonoscopy), breast cancer screening (mammogram), cervical cancer screening (smear test), osteoporosis screening (DEXA), vaccinations and diabetes examinations, such as annual foot and eye examination. Referrals (acute, symptomspecific management) included referrals to specialists. Radiology and Diagnostic Testing (acute, symptomspecific management) included all studies ordered for purposes of investigation of an acute complaint. The final category, Labs (acute and non-acute) included all laboratory testing ordered.

Using Microsoft Visio, process flow maps of several organisational processes required to complete the recommended care were generated after both direct observation of the processes and discussion with multiple staff members from the organisation.

Given the paucity of the healthcare literature found around rating process complexity, literature from the manufacturing industry provided some guidance on quantifying process complexity management. Frizelle and Woodcock[14] stated that quantification of the system can provide ‘a tool that can assist in a strategy development exercise by quantifying the problem areas on a common basis’. In 1996, Frizelle also wrote that by measuring complexity, the measurement could give ‘insight that might otherwise be missed, and allows for comparisons to be drawn’.[15] Further, Frizelle identified that there are two types of complexity: static and dynamic. Static complexity relates to the structure of the system and dynamic complexity relates to how the parts interact with each other.[15] In this study, based on suggestions for measuring complexity from the manufacturing literature[14–17] and a priori knowledge of the healthcare structure, the complexity of completing the recommended services was measured in a novel way with the following components: (1) fastest time to complete routine order (measured in weeks); (2) number of different steps leading to completion of routine order (measured by counting steps starting at order given, ending at patient arriving at specific site to complete order); (3) number of different departments involved; and (4) convenience as measured bynumber of sites visited by patient in order to complete a routine order. The first measurement was thought to be a marker of the dynamic complexity, with the remaining three measurements markers of static complexity. The sum of these four categories was then added and a composite score was calculated.

Results

During the 10-month study period, a total of 1630 recommended services and completion rates were recorded. After excluding four samples from Referrals (acute, symptom-specific services) because the documentation listed only ‘referral’, and 20 samples from Prevention and Chronic Disease Management services listed as ‘follow-up appointment’, total sample size was 1606 recommended services. This included 413 Prevention and Chronic Disease Management services, 330 Referrals (acute, symptom-specific, non-radiological, services), 180 Radiologic and Diagnostic Testing (acute, symptom-specific) and 683 Labs (acute and non-acute).

For all categories, 1111 [69% (95% adjusted Wald CI 67–71)] of the ordered services were completed without any reminder letters, 221 [14% (95% CI 12– 16)] additional ordered services were completed after one or two reminder letters and 274 (17% [95% CI 15– 19)] of the ordered services were not completed even after two reminder letters. The adherence rates were then reviewed by specific category of service by examining the effect of one reminder letter or two reminder letters (Table 1 and Figure 1). For Preventative and Chronic Disease Management services, 187 [45% (95% CI 41–50)] of the ordered services were completed without any reminder letters, 56 [14% (95% CI 11– 17)] and 28 [7% (95% CI 5–10)] were added after the first and second reminder letters, respectively, and 142 [34% (95% CI 30–39)] of the ordered services were not completed even after two reminder letters. For Referrals (acute, symptom-specific services), 219 [67% (95% CI 61–71)] of the ordered services were completed without any reminder letters, 36 [11% (95% CI 8–15)] and seven [2% (95% CI 1–4)] were added after the first and second reminder letters, respectively, and 68 [20% (95% CI 17–25)] of the ordered services were not completed even after two reminder letters. For Radiology andDiagnosticTesting (acute, symptom-specific services), 142 [79% (95% CI 72–84)] of the ordered services were completed without any reminder letters, 17 [9% (95% CI 6–15)] and two [1% (95% CI 0.04–4)] additional services were completed after one or two reminder letters, respectively, and 19 [11% (95% CI 7–16)] of the ordered services were not completed even after two reminder letters. For the Labs category, 563 [82% (95% CI 79– 85)] of the ordered services were completed without any reminder letters, 52 [8% (95% CI 6–10)] and 23 [3% (95% CI 2–5)] additional services were completed after one or two reminder letters, respectively, and 45 [7% (95% CI 5–9)] of the ordered services were not completed even after two reminder letters.

Table 1: Adherence rates for each category of services, by category of completion (baseline completion, numbers of reminder letters, and non-adherence)

Figure 1: Adherence rates for each category of services, by category of completion (baseline completion, numbers of reminder letters and non-adherence)

To understand the results further, flow maps were generated for some of the processes to delineate the patient- and system-level steps required to complete the recommended service. Figure 2a–d presents process flow maps for colonoscopies from the Preventative and Chronic Disease Management category, the general referral process to specialists from the Referrals category, the general diagnostic X-rays process from the Radiology and Diagnostic Testing category and laboratory tests from the Labs category. The process map for colonoscopy demonstrated three possible paths to completion depending on the urgency of the clinical situation, with the fastest completion for an emergency or urgent colonoscopy. The fastest routine colonoscopy completion was two weeks. The colonoscopy service also involved multiple departments including the ordering physician’s office, the gastroenterologist’s office and the medical director’s office (if deemed emergent). Process mapping for referrals to specialists demonstrated two possible paths to completion depending on the urgency of the clinical situation, and involved the ordering physician’s office aswell as the possible requirement of the patient’s participation to call for an appointment. Processmapping for general X-rays demonstrated one path to completing the X-ray (patient goes to the medical centre’s X-ray department) immediately following the ordering physician’s visit in the same building with no appointment needed. Process mapping for laboratory work demonstrated two possible paths to completion depending on the urgency (may complete some studies such as urinalysis in office) and patient’s preference to complete the laboratory tests at the medical centre or outside laboratory, with fastest completion immediately after the ordering physician’s visit at the office centre’s laboratorywithin the same building on the same floor.

Figure 2a: Colon cancer screening (colonoscopy) process flow mapping

Figure 2b: Referral to specialists process flow mapping

Figure 2c: Laboratory process flow mapping

Figure 2d: Radiographic Study (X-ray) process flow mapping

Based on process flow mapping and knowledge from the manufacturing industry literature, a novel quantitative measurement of process completion complexity was calculated. To measure the dynamic complexity, time-to-completion was examined; and, to measure static complexity, the number of departments involved, the number of locations the patient would have to visit and the number of different steps required for completing the outcome were identified. The composite scores of laboratory tests and X-rays were both calculated to be 6; referrals were calculated to have a complexity of 8; and the colonoscopy process was calculated to have the highest complexity composite score of 11–12 (Table 2).

Table 2: Process mapping complexity scores

Table 3 shows individual recommendations and completion rates, along with complexity scores for each recommendation. Some highlights of the completion rates are that, within the Preventative and Chronic Disease Management services, colon cancer screening (colonoscopy) and osteoporosis screening (DEXA) had adherence rates of 33% without any reminder letters and increased to 56% after two reminder letters. In theReferral (acute, symptomspecific) category, significant variation was seen among adherence rates across the different subspecialty referrals. In the Radiology and Diagnostic Testing (acute, symptom specific) category, several adherence rates were 100% at baseline, and many had very lownon-adherence rates.

Table 3: Adherence rates of individual recommended services, by category of completion (baseline completion, numbers of reminder letters and non-adherence), with complexity scores

Discussion

Data from this study showed similar trends with national data for patients receiving recommended services.1,18 This study further indicated, as in other published data, that patients have difficulty completing certain preventative services.1 Although the reminder letter intervention is an effective method to increase the completion rate for preventative services, 35% of patients still did not complete preventative services even after two patient reminder letters were mailed. Further research should focus on how to most effectively improve upon the reminder letter strategy or to delineate the reasons that the 35% of patients are declining even after multiple provider attempts to assist.

Comparison of the process flow complexity with adherence rates demonstrated a general trend that the more complex a process was found to be, the lower the adherence rates. For example, colonoscopy for screening purposes received a complexity score of 11–12 and was found to have one of the lowest adherence rates, even after two reminder letters; whereas, laboratory tests and general X-rays had low complexity scores with very high adherence rates. However, this inverse relationship mainly held true for non-acute categories. This suggests that healthcare complexity could be overcome in acute, symptom-specific situations. For example, completion of Referrals (acute symptom specific) had much higher adherence rates even with a moderately elevated complexity score of around 8 at this academic institution. Magnetic resonance imaging (MRI) or computed tomography (CT) scans from the acute category, estimated to have complexity scores of 8, also had very high adherence rates. Yet, mammography and DEXA scanning, with similar estimated complexity scores of 8, but from the non-acute categories, had low adherence rates. These examples suggest that acute patient needs may increase the likelihood of adherence even when complexity is high. Because the complexity composite score approach is novel, further research would need to be done in order to validate these findings and to determine complexity score thresholds, and how patient-level factors may interact with organisational complexity. Nevertheless, in the non-acute setting, it is very interesting to note that the complexity of the processes was generally found to be inversely related to the adherence rates. In future studies, process complexity may be an important system-level factor predicting adherence, which may be more important for certain patients than others.

While there was a tremendous benefit for the two reminder letters strategy, data from this study suggested that for acute, symptom-specific management (Referrals and Radiology and Diagnostic Testing) and for Labs, the reminder strategy was important but may need to be modified for these services. One reminder letter seemed to be sufficient because the second reminder letter only resulted in an additional 1–3% of recommended services completed. This echoed other published research which demonstrated the small marginal benefit from multiple reminder letters for diabetic retinopathy screening, with second reminder letters offering only increasing completions by 2%.18 For many of the recommended services in these categories, high adherence rates may be due to patients’ desire to resolve their symptoms which lead to the ordering of these recommended services initially. However, further research is needed to fully understand the differences observed and to understand if this desire for adherence may be overcome by more complex processes.

Limitations

This study has several limitations. First, patients were not surveyed to enquire reasons for declining the recommended services; future studies could be done to survey patients in order to clarify patient motivations and potential facilitators and barriers to receiving the recommended care. Second, patients with HMO insurance may have received additional reminder letters directly from the insurance carriers, which may have influenced adherence rates. Further research is needed to clarify whether the effect of managed care organisation reminders and physician–patient communication strategies work synergistically, or, possibly antagonistically, with one another. Finally, the patients in this study were from only one primary care practitioner’s outpatient office at one urban academic centre. The results from this study may not be applicable to other practices or other patient populations. However, the study did include a mix of insurance carriers and had a significant number of recommended services that may increase the generalisability of the results.

Even with the aforementioned limitations, this study did demonstrate the efficacy of the patient reminder letter strategy across multiple categories of services, and it primarily indicates the importance of using measured process flow complexity as a tool to predict adherence rates, particularly in the non-acute setting. As this finding is still in the preliminary phase, future research can more clearly investigate how measured process flow complexity can be more accurately used as a tool to predict patient health outcomes.

Acknowledgements

Presented at Society of General Internal Medicine 34^th Annual Meeting May 4^th, 2011 at Sheraton Phoenix Downtown Hotel in Phoenix, Arizona.

References

1. McGlynn EA, Asch SM, Adams J et al. The quality of health care delivered to adults in the United States. New England Journal of Medicine 2003;348(26):2635–45.
2. Jencks SF, Huff ED and Cuerdon T. Change in the quality of care delivered to Medicare beneficiaries. 1998–1999 to 2000–2001. Journal of the American Medical Association 2003;289:305–12.
3. Klabunde CN, Vernon SW, Nadel MR, Breen N, Seeff LC and Brown ML. Barriers to colorectal cancer screening: a comparison of reports from primary care physicians and average-risk adults. Medical Care 2005 Sep;43(9):939– 44.
4. Hoffman RM, Rhyne RL, Helitzer DL et al. Barriers to colorectal cancer screening: physician and general population perspectives, New Mexico, 2006. Preventing Chronic Disease 2011 Mar;8(2):A35.
5. Murakami Y, Lee BW, DuncanMet al. Racial and ethnicdisparities in adherence to glaucomafollow-up visits in a countyhospital population. Archives of Ophthalmology 2011 Jul;129(7):872–8.
6. Bastani R, Marcus AC, Maxwell AE, Das IP and Yan KX. Evaluation of an intervention to increase mammography screening in Los Angeles. Preventive Medicine 1994 Jan;23(1):83–90.
7. Richardson JL, Mondrus GT, Danley K, Deapen D and Mack T. Impact of a mailed intervention on annualmammography and physician breast examination among women at high risk of breast cancer. Cancer EpidemiologyBiomarkers and Prevention 1996 Jan;5(1):71–6.
8. Wall KM, Rocha GM, Salinas-Martinez AM, Baraniuk S and Day RS. Modifiable barriers to cervical cancer screening adherence among working women in Mexico. Journal of Women’s Health (Larchmt) 2010 Jul;19(7): 1263–70.
9. DeVoe JE, Baez A, Angier H, Krois L, Edlund C and Carney PA. Insurance+access6¼health care: typology of barriers to health care access for low-incomefamilies. Annals of Family Medicine 2007;5:511–18.
10. Committee on Quality of Health Care in America, Institute of Medicine. Crossing the QualityChasm: a new health system for the 21st century. National AcademiesPress: Washington DC, 2001.
11. Rosser WW, Hutchinson BG, McDowell I and Newell C. Use of reminder to increase compliance with tetanusbooster vaccination. CanadianMedical Association Journal 1992;164(6):911–17.
12. Winston CA, Mims AD and Leatherwood KA. Increasing pneumococcal vaccination in managed care throughtelephoneoutreach. American Journal of Managed Care 2007;13:581–8.
13. Kaczorowski J, Karwalajtys T, Lohfeld L et al. Women’sviews on reminderletters for screening mammography. Canadian FamilyPhysician 2009;55:622–3. e1–4.
14. Frizelle G and Woodcock E. Measuringcomplexity as an aid to developing operationalstrategy. International Journal of Operations & Production Management 1995; 15(5)5:26–39.
15. Frizelle G. Getting the measure of complexity. ManufacturingEngineer 1996;75(6):268–70.
16. Pippenger N. Complexitytheory. Scientific American 1978 Jun;238(6): 114–25.
17. Klir G. Complexity: some general observations. Systems Research 1985;2:131–40.
18. Halbert RJ, Leung KM, Nichol JM and Legorreta AP. Effect of multiple patient reminders in improving diabeticretinopathy screening. A randomized trial. Diabetes Care 1999 May;22(5):752–5.