Response Rates of Three Modes of Survey Administration and Survey Preferences of Rheumatoid Arthritis Patients

Authors


Abstract

Objective

To compare survey response rates and preferences for 3 modes of survey administration in rheumatoid arthritis (RA) patients.

Methods

Adult RA patients were identified from Kaiser Permanente Southern California's electronic medical records. One hundred patients each were randomly assigned to telephone, mail with followup letter, and mail with followup telephone call as modes of survey administration. Respondents completed a 7-page survey (in English or Spanish) that included the Health Assessment Questionnaire disability index and pain scale, EuroQol 5-dimensions, and generic questions on preferences and sociodemographics. Response rates were based on the number of completed surveys received from the total number of patients contacted (n = 295; 5 were ineligible).

Results

The mean ± SD age of the cohort was 61 ± 12.3 years, with the majority being women (141 [85%] of 166). Of the 3 modes of survey administration, telephone (63 [64%] of 99) had the highest response rate, followed by mail with followup telephone call (55 [56%] of 98) and mail with followup letter (48 [49%] of 98). When asked about preference over administration mode, 57% (95 of 166) preferred to complete the survey by mail, followed by telephone (27 [16%] of 166) and internet (17 [10%] of 166).

Conclusion

When asked about their choice of survey administration mode for reporting their health status, the majority of RA patients in this study had a strong preference for a mail survey as compared to a telephone interview. However, the response rate in the telephone administration mode of survey was the highest between the 3 modes of administration that were compared. Future studies in RA can achieve better response using telephone over mail-in surveys.

INTRODUCTION

Rheumatoid arthritis (RA) is an autoimmune disease that causes erosive synovitis resulting in joint damage and loss of joint function, leading to early disability and decreased functional status ([1]). Women are 2–4 times more likely than men to develop RA ([2]). Besides sex, RA is rising in the elderly. Between 1995 and 2007, the incidence rate of RA increased with age, at 8.7 per 100,000 among those ages 18–34 years compared with 54 per 100,000 among those ages ≥85 years; incidence peaked among people ages 65–74 years (89 per 100,000) ([3, 4]). RA currently affects approximately 1.3 million adults in the US; however, over the next 25 years, the number of people affected and the social impact of doctor-diagnosed arthritis are projected to increase by 40% ([1]).

One of the primary data collection instruments in social and epidemiologic research is the survey questionnaire ([5]). Most rheumatologists rely on their patient's own self-report of their condition, a new gold standard for RA outcomes, in addition to laboratory and physician measurements ([6]). Patient-reported outcomes of physical function/disability, such as those reported by the Health Assessment Questionnaire (HAQ) and the Short Form 36 health survey, are the most widely used fundamental outcome measures in rheumatology clinical studies ([7]). Modes of data collection by questionnaire vary in the method of contacting respondents, the delivery of the questionnaire, and the format in which questions are administered ([5]). Survey data collection can be carried out by in-person interviews, telephone interviews, mail questionnaires, or a combination of methods ([8]). Collection of self-reported survey data among patients with RA can differ from the general population due to the disability associated with this disease.

The influence of physical disability and cognitive dysfunction on the response rate is important because these disabilities make it more difficult to measure health-related quality of life in RA patients ([9]). RA has a propensity to target the wrist and the metacarpophalangeal joints and proximal interphalangeal joints of the hand, resulting in 90% of individuals developing hand deformities during the course of their disease. These deformities have significant functional consequences and impact quality of life because they impair hand strength, dexterity, mobility, and the ability to use hands efficiently for daily activities ([10]). The most burdensome modes of survey participation for RA patients are likely to be written and visual methods of self-administration because these demand that respondents be literate in reading the language(s) of the survey, that they do not have visual impairments, and that they have the dexterity (e.g., of the wrists and fingers) to complete the survey ([5]). Studies have shown that survey response rates are related to age, disability, and cognitive impairment; therefore, those with poor physical and mental health have difficulty in returning self-completed questionnaires ([9, 11]). These challenges are improved upon by survey administration with a trained data collector ([9]).

Unfortunately, trends in RA show that the incidence of disease increases with age, peaking between the fourth and sixth decade, which presents challenges when conducting surveys in this population ([9, 12]). In elderly RA patients, the reasons for physical disability are even more complex, since the decline of physical function related to aging must be considered ([12]). Low response rates and problems with missing data have been noted in ambulatory care, particularly in the very elderly, compared to older people who are at home and are more likely to be able to complete and return the questionnaire ([9]).

When focusing on the modes of survey administration, studies comparing mail surveys to other methods of data collection, such as telephone surveys and in-home interviews, indicate that mail surveys have lower response rates. Poor response rates to mailed surveys have been credited to multiple reasons: unattractive, unclear or offensive content or design, holiday seasons, language, type of return envelope and postage used, study patients forgetting to return the questionnaire, lack of patient incentives, length of questionnaires, an uninteresting study, and lack of mail or phone reminders, among others ([13]). An advantage of mail surveys is they do allow more anonymity for reporting personal and sensitive information such as morbidity, disability, and socially undesirable behaviors ([14]). The single greatest advantage of mailed self-administered questionnaires is their lower cost compared with other methods (e.g., in-person and telephone interviews). A completed questionnaire administered by mail costs approximately 50% less than one administered by telephone, and 75% less than in-person interviews ([15]).

Nonresponse to a survey is likely to be influenced by the length of the survey or the mode of administration (e.g., people who have difficulty writing are unlikely to respond to a postal survey) ([5, 8]). Evidence exists that response rates to mailed surveys may be increased through reminder phone calls to nonresponders ([13, 15]). Mailed surveys with telephone interview followup is better than telephone interview with mail followup ([5]). Response rates can also be improved by the personal in-person survey administration, which is the least burdensome method and only requires the respondent to speak the same language in which the questions are asked and to have basic verbal and listening skills ([5, 9]).

Given the abovementioned challenges in collection of the critical patient-reported outcomes in RA, the aim of this study was to learn about successful forms of collecting self-reported data based on the response rates to surveys from a diverse RA population in a large integrated care organization. The study allowed us to compare the effectiveness of several survey data collection methods and to learn about RA patients' preferences in participating in survey data collection and the heterogeneity in such preferences.

Box 1. Significance & Innovations

  • Collection of patient-reported outcomes data about disease status among rheumatoid arthritis (RA) patients is considered a gold standard for assessing health outcomes; however, challenges for RA patients are different from the general population due to the physical disability associated with this disease.
  • In this study, we evaluated the effectiveness of 3 distinct modes of survey administration (mail with letter followup, mail with telephone followup, and telephone only) to collect patient-reported outcomes as well as identify patient preferences for each of these survey administration modes among RA patients in a large integrated care organization.
  • The highest survey response rate was observed in the telephone mode, followed by mail with followup telephone call and mail with followup letter. Patients indicated mail surveys as the most preferred mode of survey administration.

PATIENTS AND METHODS

Study population

The study population included members from the Kaiser Permanente Southern California (KPSC) managed care health plan who were diagnosed with RA. KPSC is an integrated health care system that provides comprehensive health services to approximately 3.5 million residents of Southern California. The population served by KPSC is socioeconomically diverse and broadly representative of the racial/ethnic groups living in Southern California ([16]). Members enroll through the Kaiser Foundation Health Plan for prepaid health care insurance, including pharmaceutical benefits. The KPSC region includes 14 hospitals and more than 202 medical offices and a partnership of over 5,800 physicians who comprise the entire range of medical specialists.

Data source

The primary data source was KPSC electronic medical records (EMRs), which contain a detailed account of interactions of members with the health care system. Disease-modifying antirheumatic drug (DMARD) prescription data were obtained from the KPSC EMR, while patient demographic and health plan enrollment information was obtained from KPSC membership databases.

Study cohort and study design

This prospective cohort study was approved by the KPSC Institutional Review Board. Three hundred RA patients between the ages of 18 and 100 years with a clinical diagnosis of RA were identified from the EMR between November 1, 2010 and December 13, 2010, and were prospectively followed over time at baseline and 3- and 6-month intervals. For inclusion into the study, patients were required to have ≥2 medical records separated by 1 month, with a primary or nonprimary diagnosis of RA as defined by International Classification of Diseases, Ninth Revision, Clinical Modification codes in the year prior to the use of a DMARD prescription for RA. Pregnant women, as well as patients who were cognitively impaired, were excluded from the study because their quality of life, working ability, and productivity would be heavily influenced by their condition.

At baseline, patients were randomly assigned to one of the following cohorts as the mode of administration: telephone, mail with reminder followup letter, and mail with reminder followup telephone call (Figure 1).

Figure 1.

Study illustration and response rate calculations. RA = rheumatoid arthritis.

The patients received information about the study rationale and objectives either by letter or telephone. Only after a subject voluntarily agreed to participate was he/she recruited in the study. Patients were given the option to decline participation, and those who received mail surveys were also sent an “opt-out” postcard should they choose not to participate. Return of completed surveys by mail or response to the survey questions over the telephone implied consent to participate in the research study. All patients were invited to complete the survey, which included the HAQ disability index (DI) and pain scale, EuroQol 5-dimensions (EQ-5D), and generic questions on sociodemographics and preferences. The HAQ DI is designed to measure functional status (where higher scores represent higher disability), while the EQ-5D is a generic health-related quality of life and health utility elicitation instrument (scores range from −0.11 for the lowest health state to 1.0 for full health for the US population). The survey took approximately 20 minutes to complete and was self-administered for the mail cohort. The study patients participated voluntarily and were not offered an incentive (e.g., monetary) to participate in this study. The survey was provided in either English or Spanish based on the patient's preference. Patients were not included in the study if they could not understand English or Spanish.

Patients in the mail cohort were contacted with a reminder followup letter or phone call if a completed survey was not received within 2 weeks of the original mailing date. If an opt-out card was received or if the patient did not complete the survey within 1 month after the original date it was mailed, patients were considered to have declined participation in the study. Patients in the telephone cohort were attempted to be reached at different times during the day. Research staff was flexible with making calls to potential study patients by calling in the late evening and on weekends. A telephone patient was considered to decline participation when 3 voicemails were left on 3 distinct days approximately 1 week apart or when a patient missed more than 2 scheduled call-back appointments.

Outcomes

The primary outcome for this study was the comparison of the response rates at baseline. Additionally, we also evaluated patients' preferences for the different modes of survey administration, allowing us to compare and contrast the effectiveness of and preferences for each mode of administration. Advantages and disadvantages of different modes of survey administration among the RA patients were gathered from the observations of the research staff. Differences by age groups and racial/ethnic differences in preferences as well as differences in response rates based on language of survey administration were also evaluated.

Statistical analysis

Descriptive statistics, including means, medians, and proportions, were used to understand data distribution. Differences between proportions were assessed by chi-square tests, while differences between means were evaluated by analysis of variance and Student's t-test. Response rates were calculated based on the number of completed surveys received from the total number of patients contacted.

RESULTS

A total of 300 patients were randomly assigned to 1 of the 3 possible data collection cohorts. However, upon patient medical chart audit, it was determined that 5 patients were ineligible for the following reasons: no longer an active member of the KPSC health care plan (n = 1), deceased (n = 2), or no longer had a diagnosis of RA (n = 2). This resulted in a reduced total starting cohort of 295. Specifically, within each cohort, we contacted 99 patients by telephone, 98 by mail with a reminder followup letter, and 98 by mail with a reminder followup telephone call.

Of the 295 surveys that were administered, only 166 surveys were completed, with an overall response rate for all modes of administration of 56%. Of the 3 modes of administration, the telephone cohort had the highest response rate (63 [64%] of 99), followed by mail with a followup telephone call (55 [56%] of 98) and mail with a followup letter (48 [49%] of 98) (Figure 1). The combined overall response rate for both mail groups was 53%.

Among those who completed the survey (n = 166), the majority was women (141 [85%] of 166). The mean ± SD age was 61.0 ± 12.3 years and the mean ± SD duration of RA was 12.8 ± 11.4 years (Table 1). The majority of the patients in all 3 of the randomized cohorts were white: telephone (52%), mail with a followup letter (53%), and mail with a followup telephone call (45%) (Table 1). The mean ± SD HAQ DI score was 1.0 ± 0.70 and the mean ± SD EQ-5D score was 0.70 ± 0.20.

Table 1. Descriptive patient characteristics*
 Telephone (n = 63)aMail with followup call (n = 55)aMail with followup letter (n = 48)aOverall (n = 166)
  1. Values are the number (percentage) unless indicated otherwise. HAQ = Health Assessment Questionnaire; DI = disability index; EQ-5D = EuroQol 5-dimensions; RA = rheumatoid arthritis.
  2. aDifferences between the cohorts were not statistically significant at an alpha level of 0.05 for any of these variables.
  3. bRange 0–100, where 0 = no pain and 100 = severe pain.
  4. cRange 0–100, where 0 = very well and 100 = very poor health.
Age, mean ± SD years59.7 ± 12.061.3 ± 12.161.9 ± 13.160.9 ± 12.3
HAQ DI score, mean ± SD1.0 ± 0.80.9 ± 0.70.9 ± 0.71.0 ± 0.7
EQ-5D utility score, mean ± SD0.7 ± 0.20.7 ± 0.20.7 ± 0.20.7 ± 0.2
Self-rated pain, mean ± SDb48.9 ± 31.638.0 ± 27.741.7 ± 28.043.2 ± 29.5
Self-rated health, mean ± SDc39.4 ± 29.052.7 ± 32.347 ± 29.146.0 ± 30.5
Duration of RA, mean ± SD years13.0 ± 11.411.8 ± 10.813.6 ± 12.212.8 ± 11.4
Female sex55 (87.3)46 (83.6)40 (83.3)141 (84.9)
Race/ethnicity (n = 6 missing)    
White32 (51.6)24 (45.3)24 (53.3)80 (50)
African American6 (9.7)6 (11.3)3 (6.7)15 (9.4)
Hispanic/Latino22 (35.4)20 (37.7)15 (33.3)57 (35.6)
Asian/Pacific Islander1 (1.6)3 (5.7)3 (6.7)7 (4.4)
Other1 (1.6)0 (0.0)0 (0.0)1 (0.6)
Language    
English44 (69.8)39 (70.9)40 (83.3)123 (74.1)
Spanish18 (28.6)13 (23.6)8 (16.7)39 (23.5)
Other1 (1.6)3 (5.5)0 (0)4 (2.4)
Level of education (n = 4 missing)    
No college degree43 (69.4)33 (63.5)27 (56.3)103 (63.6)
College degree19 (30.6)19 (36.5)21 (43.7)59 (36.4)
Occupation (n = 3 missing)    
Working for pay25 (39.7)18 (34.0)18 (38.3)61 (37.4)
Retired19 (30.2)23 (43.4)21 (44.7)63 (38.7)
Disabled and not working8 (12.7)5 (9.4)5 (10.6)18 (11.0)
Homemaker10 (15.9)5 (9.4)2 (4.3)17 (10.4)
Looking for work1 (1.6)2 (3.7)1 (2.1)4 (2.5)
Lives alone (n = 3 missing)8 (12.9)11 (20.8)8 (16.7)27 (16.6)

When asked about their preference on the mode of survey administration, more than half (57%) of the study patients would have preferred to complete the survey by mail, followed by telephone (16%) and the internet (10%) (Table 2). Even with random assignment, the mailing mode of administration was highly preferred by those who had participated in the survey by mail (76%). However, the telephone cohort preferred to complete the survey by telephone (40%), followed by mail (27%). None of the cohorts showed a preference to completing the survey in person (Table 2).

Table 2. Patient-stated survey preference of survey administration
 Survey administration cohort, no. (%)
Telephone (n = 63)aMail with followup call (n = 55)aMail with followup letter (n = 48)aOverall (n = 166)
  1. aDifferences between the cohorts were not statistically significant at an alpha level of 0.05 for any of these variables.
Stated survey preference    
Mail17 (27.0)43 (78.2)35 (72.9)95 (57.2)
Telephone25 (39.7)1 (1.8)1 (2.1)27 (16.3)
In person0 (0)0 (0)0 (0)0 (0)
Internet8 (12.7)5 (9.1)4 (8.3)17 (10.2)
Other12 (19.1)3 (5.5)7 (14.6)22 (13.3)
Missing1 (1.6)3 (5.5)1 (2.1)5 (3.0)

Choice of the survey administration mode varied by respondents' language preferences and not by age or race (Table 3). The majority of patients ages >65 years (68%) as well as those ages ≤65 years (51%) preferred to participate in the survey by mail, followed by telephone (15% and 17%, respectively) (Table 3). However, respondents ages ≤65 years had a higher preference to completing the survey via the internet (7% and 13%, respectively).

Table 3. Stated survey preferences by age, language, and race*
 MailTelephoneInternetOtherMissingTotal, no.
  1. Values are the number (percentage) unless indicated otherwise.
  2. aDifferences in stated preferences by language were statistically significant at an alpha level of 0.05.
Age, years      
≤6553 (51.0)18 (17.3)13 (12.5)16 (15.4)4 (3.9)104
>6542 (67.7)9 (14.5)4 (6.5)6 (9.7)1 (1.6)62
Race (n = 4 missing)      
White42 (52.5)13 (16.3)10 (12.5)15 (18.8)0 (0)80
Nonwhite51 (62.2)14 (17.1)7 (8.5)7 (8.5)3 (3.7)82
Languagea      
English72 (58.5)17 (13.8)16 (13.0)18 (14.6)0 (0)123
Spanish/bilingual21 (53.9)10 (25.6)1 (2.6)4 (10.3)3 (7.7)39
Other2 (50.0)0 (0)0 (0)0 (0)2 (50.0)4

The preferences between whites and nonwhites in the survey administration followed a similar trend, with mail being the most favored method, followed by telephone (16% and 17% in whites and nonwhites, respectively) and the internet (13% and 9% in whites and nonwhites, respectively), for the provided options. Of those patients (n = 95) who preferred the mail mode of survey participation, most were nonwhite (n = 51 [53.7%]). Of the 17 patients who preferred the internet mode of survey participation, more than half (n = 10 [59%]) were white.

The majority of Spanish-speaking/bilingual (English- and Spanish-speaking) patients (n = 21) preferred to participate by mail (54%). Of those who preferred to participate in the study via the internet (n = 17), the majority were English speaking (n = 16 [94%]). English-speaking patients equally preferred to participate in the survey via the internet (13%) and telephone (14%). This differed from Spanish-speaking/bilingual patients in that they preferred to complete the survey via telephone (26%) or over the internet (3%) (Table 3).

Overall, for the demographic factors we examined, it appeared that younger, white, English speakers preferred to use the internet compared to older, nonwhite, non–English speakers. Those ages ≤65 years were 2.5 times as likely to prefer using the internet as those ages >65 years, whites were 1.4 times as likely to prefer using the internet over nonwhites, and English speakers were 16 times as likely to prefer the internet as compared to non–English speakers (Table 3).

Of the 166 RA patients who completed the survey, 141 (85%) agreed to followup survey rounds within 3 months (Table 4). Overall, the study patients seemed willing to participate in future surveys even with additional questions, with 75% agreeing to a survey that would take 30 additional minutes (Table 4). The mail cohorts, however, seemed more open to the idea of filling out additional questions on a survey (mail with a followup telephone call [80%] and mail with a followup letter [77%]). The telephone cohort (70%) was the least willing to participate in subsequent rounds of surveys (Table 4). More than half (59%) of the overall patients agreed that reimbursement would encourage them to participate in the study (Table 4). Of the 141 patients who had agreed to a followup survey, only 113 (80%) completed the 3-month followup survey, resulting in an attrition rate of 20%.

Table 4. Willingness to participate on additional surveys*
 Telephone (n = 63)aMail with followup call (n = 55)aMail with followup letter (n = 48)aOverall (n = 166)
  1. Values are the number (percentage).
  2. aDifferences between the cohorts were not statistically significant at an alpha level of 0.05 for any of these variables.
  3. bOnly applicable to those who answered yes to the previous question.
If we follow up with the same questions in 3 months and 6 months, will you still participate?    
Yes58 (92.1)42 (76.4)41 (85.4)141 (84.9)
No5 (7.9)12 (21.8)6 (12.5)23 (13.9)
If additional questions are added to this survey that required 30 additional minutes, will you still participate?    
Yes44 (69.8)44 (80.0)37 (77.1)125 (75.3)
No18 (28.6)11 (20.0)10 (20.8)39 (23.5)
Will a reasonable reimbursement of your time encourage you to participate?    
Yes40 (63.5)27 (49.1)31 (64.6)98 (59.0)
No22 (34.9)25 (45.5)15 (31.3)62 (37.4)
Missing1 (1.6)3 (5.5)2 (4.2)6 (3.6)
Please provide a reasonable reimbursement amount for your participation in each survey.b    
$2516 (40.0)8 (29.6)12 (38.7)36 (36.7)
$506 (15.0)4 (14.8)2 (6.5)12 (12.2)
$750 (0)1 (3.7)6 (19.4)7 (7.1)
$1004 (10.0)7 (25.9)6 (19.4)17 (17.3)
Other13 (32.5)2 (7.4)4 (12.9)19 (19.4)
Missing1 (2.5)5 (18.5)1 (3.2)7 (7.1)

DISCUSSION

To our knowledge, no other studies have focused on understanding barriers to collecting patient-reported outcomes by comparing survey response rates and survey participation preferences among patients with RA. Using the KPSC population, we were able to recruit a random sample of RA patients belonging to a large and racially diverse integrated care health plan. Our study provides comparative effectiveness and preferences among 3 of the most commonly used approaches for collecting patient-reported outcomes, which are crucial for management of RA patients. The strengths of our study include the prospective cohort design and the large sample size with random assignment to the different modes of survey administration. We also longitudinally followed up with the patients to understand the dropout rate over time. The disease status of each patient was confirmed by reviewing their medical charts. Our results on patient preferences of participation can inform physicians and researchers about which preferred modes of survey administration are likely to have higher participation rates among RA patients.

Our key finding is that “revealed/true” response rates among the 3 cohorts were not indicative of their “stated/self-reported” preference of mode of survey administration among RA patients. The telephone cohort had the highest response rate (64%), but the majority of these patients had stated a preference for mail surveys. The higher response rates in the telephone and mail with a followup telephone call modes of survey administration could be attributed to the direct contact with our research staff in these 2 modes.

A majority of the study patients were also willing to participate in a followup survey. This may be due to the fact that patients with health problems related to the disease of interest are more motivated to participate and be involved in research concerning “their disease” ([17]). Also, many reported that a reasonable reimbursement for their time would encourage them to participate in followup surveys, with $25 being indicated as an acceptable reimbursement amount.

The preference of survey administration among patients with RA may differ from the general population due to functional impairment and disability, which are primary manifestations of RA ([18]). Therefore, completion of a survey can be more demanding on a patient with RA, although its output is very important in understanding their disease status, severity, and progression. Bowling showed higher response rates for self-administered postal questionnaires compared with self-administered questionnaires handed out to people (e.g., in hospitals) to complete and return ([5]). In contrast, our RA population had a higher response rate for telephone-administered questionnaires compared to those that were mailed. However, RA patients in this study preferred self-administered mail surveys over the telephone surveys. The greater anonymity offered in a postal survey, with its weak social presence, has been reported to lead to high item response and more accurate reporting on sensitive topics, such as health and behavior ([5]).

Since RA incidence increases with age, the attrition rate of 20% could be related to the fact that response rates usually decrease with age and decreases in those with lower socioeconomic status ([9, 11, 12]). Women are most commonly affected by RA, which aids in increased response rates because women are more likely than men to participate in scientific studies ([2, 19]).

Nonwhites greatly preferred mail administration of the survey and Spanish speakers also preferred to participate in the survey by mail. Patients ages >65 years did not have a strong preference to completing the survey on the internet compared to those ages ≤65 years. This may be due to the older generation having less experience with the internet. Older RA patients were not born in an era when the internet existed. They had to learn how to use the internet when it first appeared; however, younger RA patients have been using the internet since a young age. Whites were slightly more open to the idea of completing the survey over the internet. Furthermore, availability of the internet, especially in remote areas, may also be a factor affecting this preference.

Issues of the missing data in surveys and incorrect filling out of surveys were seen in the mail self-administered surveys; nonetheless, it was the preferred mode of administration. Mail surveys were the least expensive mode of administration, and patients were able to complete the survey at a time that was most convenient to them. Between the 2 modes of mail administration, telephone followup was more beneficial than mail followup in order to increase the response rate.

Finding a convenient time to complete the survey proved most difficult for telephone patients. Telephone surveys often required that multiple voice messages be left and call-back appointment times scheduled, and research staff worked evenings and weekends to reach the study participants. Study patients also sometimes missed scheduled call-back appointments. One advantage of the telephone surveys, however, was that they were completed properly since they were administered by the research staff.

This study has important implications to the methodology of future questionnaire studies with the RA population. The 3 different forms of followup with RA patients allowed us to recruit more people. RA patients who preferred to complete the survey by telephone were still able to do so via followup phone calls. Overall, the majority of RA patients preferred to participate in a self-administered survey rather than one involving the research staff; however, higher involvement of the research staff produced a better response rate. One of the limitations of the study is that a patient's willingness to participate in a survey may differ depending on if they are experiencing a disease flare or a period of worsening disease activity. If patients are currently experiencing a disease flare, their willingness to participate in the survey may be significantly decreased because they are in pain and prefer not to be bothered. We did not control for disease activity at the time of survey completion. Future studies could aim to focus on response rate and survey administration preferences among RA patients experiencing a disease flare.

In conclusion, this study showed that the unique disability characteristics of RA should be addressed when choosing the survey mode of administration. RA patients' preference of self-administered mail survey should be taken into consideration. For future studies, we recommend multiple modes of survey administration to maximize recruitment and response rates. This would allow for a maximum number of RA patients to be reached based on the reported preference of survey administration. If only one administration mode is feasible, then we recommend telephone survey administration whenever possible. Lastly, between the 2 mail survey options, followup by a reminder telephone call with the option for the patient to complete the survey over the phone is preferable over reminder mail or postcard.

AUTHOR CONTRIBUTIONS

All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Kawatkar had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. Chu, Kawatkar.

Acquisition of data. Garcia, Portugal, Chu, Kawatkar.

Analysis and interpretation of data. Garcia, Chu, Kawatkar.

Acknowledgments

The authors would like to thank Mr. Rajan Iyer for his contributions to the data analysis in support of the revised manuscript.

Ancillary