Factors associated with phlebotomy failure in an Alzheimer's Disease Research Center cohort

Abstract

INTRODUCTION

Venous blood collection through phlebotomy is critical for biomarker development and genetics research on Alzheimer's disease (AD) and related dementias. In this research setting, phlebotomy is often performed by clinical research coordinators (CRCs) rather than professional nurses or phlebotomists. Even though phlebotomy in older adults with cognitive impairment can be empirically challenging, data on the factors associated with phlebotomy failure in this special population are lacking. We addressed this question in the Massachusetts Alzheimer's Disease Research Center (MADRC) Longitudinal Cohort.

METHODS

We applied mixed‐effects logistic regression with phlebotomy failure as the dependent variable; age at visit, sex, education, body mass index (BMI), under‐represented minority (URM) status, and Clinical Dementia Rating (CDR) scale global score as fixed effects; and participant ID as a random effect to account for multiple visits per participant.

RESULTS

Full data were available for a total of 1365 visits from 659 individual participants (average 2.1 visits/participant, range 1–7). Greater BMI (odds ratio [95% confidence interval]: 1.10 [1.05–1.16], P < 0.001) and CDR global score (1.51 [1.13–2.03], P = 0.005), but not age, sex, or education, were independently associated with higher odds of phlebotomy failure. URM participants had twice the odds of a phlebotomy failure compared to non‐Hispanic Whites, although this trend did not reach statistical significance due to insufficient power (2.07 [0.97–4.41], P = 0.059).

DISCUSSION

Greater BMI and CDR as well as URM status are independently associated with higher odds of phlebotomy failure. Potential strategies to maximize participation of these subpopulations in biomarker and genetics research are discussed.

Highlights

• We analyzed factors underlying failed phlebotomy in the Massachusetts Alzheimer's Disease Research Center cohort.

• Higher body mass index and Clinical Dementia Rating global score were associated with phlebotomy failure.

• Under‐represented minorities had twice the odds of suffering failed phlebotomy compared to non‐Hispanic Whites.

1. INTRODUCTION

Blood‐based biomarkers are being adopted in clinical practice as a tool for the early diagnosis, differential diagnosis, biological staging, prognostication, and therapeutic decision making of Alzheimer's disease (AD) and AD related dementias (ADRD). ¹ , ² In observational research studies, plasma phosphorylated tau 217 (p‐tau217) is often used in conjunction with neuropsychological testing and functional scales (e.g., Clinical Dementia Rating [CDR] scale) to stage participants as preclinical, subjective cognitive decline, mild cognitive impairment, and mild/moderate/severe dementia due to AD. Additionally, blood collection is critical for biomarker development ³ , ⁴ and genetics research. ⁵ In this setting, phlebotomy is often performed by clinical research coordinators (CRCs) or research assistants rather than professional phlebotomists or nurses with years of experience. ⁶ Although phlebotomy in older adults may be challenging due to frailty and cognitive impairment, data on the factors associated with phlebotomy success/failure in this research setting are limited. ⁷ Yet, any demographic or clinical factor associated with phlebotomy failure could introduce a systematic selection bias leading to under‐representation of a specific population in biomarker and genetic research studies.

We sought to examine the factors associated with phlebotomy failure in the Massachusetts Alzheimer's Disease Research Center (MADRC) Longitudinal Cohort Study.

2. METHODS

2.1. Study participants and data collection

The MADRC Longitudinal Cohort Study is part of the National Institute on Aging (NIA)‐funded National Alzheimer's Coordinating Center (NACC) longitudinal cohort study, which encompasses more than 35 Alzheimer's Disease Research Centers in the US. ⁸ , ⁹ Participants are recruited from the Massachusetts General Hospital (MGH) neurology clinics and from the community through community events, advertisements in social media, and the MADRC website (www.madrc.org). All participants and/or their chosen knowledgeable informant (usually the spouse or next of kin) agreed to participate in the MADRC Longitudinal Cohort Study, which is approved by the MGH Institutional Review Board (IRB; protocol #1999P003693). After written informed consent, participants undergo annual visits with standardized questionnaires and forms, anthropometrics (weight, height, body mass index [BMI]), vital signs, neurological exam, and neuropsychological test battery, collectively known as the Uniform Data Set (UDS). ⁸ , ⁹ The following variables were selected from the UDS: visit age, sex, education, race/ethnicity, BMI, and Clinical Dementia Rating (CDR) global score obtained with the CDR^® Dementia Staging Instrument. ¹⁰ Age and BMI were treated as continuous variables, whereas education was dichotomized between ≤ 12 and > 12 years, and race and ethnicity were combined and dichotomized between non‐Hispanic Whites and under‐represented minority (URM) populations.

2.2. Eligibility criteria

Only participants with attempted phlebotomy between 2015 and 2024 were included in this study. Participants’ visits in which phlebotomy was not attempted (for example, due to remote/virtual visit or participant's refusal) or in which information about phlebotomy was not collected were excluded.

RESEARCH IN CONTEXT

1. Systematic review: We reviewed the English and Spanish literature on predictors of phlebotomy failure using PubMed with no search filters. We found multiple papers addressing this question in the inpatient setting (e.g., critically ill or surgical patients) and in the context of nursing school training, but none focused on a vulnerable research population such as older adults with cognitive impairment.

2. Interpretation: Our analysis in the Massachusetts Alzheimer's Disease Research Center Longitudinal Cohort Study identified higher body mass index and Clinical Dementia Rating scale global score, and self‐reported under‐represented minority status as independent factors associated with increased odds of phlebotomy failure by clinical research coordinators.

3. Future directions: We discuss several possible strategies to maximize clinical research coordinators’ phlebotomy success rate in these subpopulations of research participants.

2.3. Phlebotomy procedures

At every visit, participants are asked to provide a blood sample for biomarker and genetics research. All participants and/or their next of kin signed provided a separate written consent form to agree to phlebotomy and the research use of their blood sample under the MGH IRB‐approved protocol #2006P002104.

Blood is drawn via phlebotomy in the antecubital vein by CRCs who are typically hired after completing their bachelor's degree, have no prior phlebotomy experience, and remain in the team for 1 to 3 years before pursuing new career endeavors. Phlebotomy training for newly hired CRCs is standardized: First, they complete the Mass General Brigham E‐learn HealthStream Phlebotomy online course; then, they attend a 1‐hour hands‐on session led by an MGH registered nurse or phlebotomist in which they practice with a phlebotomy simulation arm and can practice on another CRC attendee. Last, they complete six practice venipunctures on peer volunteers before being allowed to conduct phlebotomies on research participants.

The CRCs record the outcome of phlebotomy at every visit in a MADRC local Research Electronic Data Capture (REDCap) form as either successful or not completed, with reasons for the latter further classified as: (1) difficult veins (i.e., phlebotomy failure), (2) remote/virtual or phone visit, (3) participant's refusal, (4) medical problem, and (5) other reasons (e.g., visit time constraints).

2.4. Statistical analyses

All statistical analyses were conducted using STATA v15.0 (StataCorp, LLC, College Station, TX). Univariate analyses compared participants’ visits with successful phlebotomy versus failed phlebotomy due to difficult veins. Normality of continuous variables was tested with the Shapiro–Wilk, kurtosis, and skewness tests, as well as inspection of distribution histograms. Because continuous variables (i.e., age at visit, years of education, and BMI) were non‐normally distributed, we ran Wilcoxon rank‐sum tests for univariate comparisons between the successful and failed phlebotomy groups. For categorical variables (i.e., sex, education ≤ 12 and > 12 years, URM status, and CDR global score), we tested differences in proportions with a chi‐squared test.

To investigate the factors independently associated with phlebotomy failure, we applied mixed‐effects logistic regression models with phlebotomy success/failure as the binary outcome variable; age, sex, education, URM status, BMI, and CDR global score as fixed effects; and participant's NACC ID as a random effect to control for within‐participant correlation of phlebotomy attempts across multiple visits, as follows:

$$phlebotomy\sim age+sex+education+urm+bmi+cdr\parallel nacc\_id$$

3. RESULTS

We analyzed a total of 1491 in‐person visits from 727 participants between 2015 and 2024. Of these, the CRCs failed to complete the phlebotomy due to difficult veins in 184 (12.3%). Table 1 shows the demographic and clinical characteristics of study participants by visits with successful versus failed phlebotomy and the comparisons between visits with both phlebotomy outcomes. There were 188 URM participants, including 76 self‐identified as Hispanic/Latino, 111 as non‐Hispanic Black, and one as other (Cape Verdean). Univariate analyses revealed that participants in visits with failed phlebotomy had lower education, were twice as likely to self‐identify as URM, had greater BMI, and were more impaired based on CDR global score than those with successful phlebotomy. In contrast, the two groups did not significantly differ in age at visit or sex.

TABLE 1.

Baseline demographic and clinical characteristics of study participants.

	Total	Phlebotomy		p value
		Success	Failure
Number of visits	1491	1307 (87.7)	184 (12.3)	N/A
Visit age (years)	74.0 ± 11.6	74.0 ± 11.5	74.0 ± 11.9	0.955
Sex, n (%) female	852 (57.1)	743 (56.9)	109 (59.2)	0.539
Education (years)	16.4 ± 2.7	16.4 ± 2.7	15.8 ± 2.8	0.001
Education > 12 years, n (%)	1301 (87.5)	1150 (88.2)	151 (82.5)	0.030
URM, n (%)	188 (12.6)	147 (11.3)	41 (22.3)	<0.001
BMI (kg/m2)	26.8 ± 5.0	26.5 ± 4.7	29.0 ± 6.2	<0.001
CDR global score, n (%):				0.001
0.0	501 (33.9)	454 (35.1)	47 (25.7)
0.5	767 (51.9)	673 (52.0)	94 (51.4)
1.0	150 (10.2)	124 (9.6)	26 (14.2)
2.0	46 (3.1)	33 (2.6)	13 (7.1)
3.0	14 (1.0)	11 (0.9)	3 (1.6)

Notes: Continuous variables are expressed as mean ± standard deviation and were compared with a Wilcoxon rank‐sum test. Categorical variables are shown as column % and were compared with a chi‐squared test. Statistically significant results are in bold font.

Abbreviations: BMI, body mass index; CDR, Clinical Dementia Rating scale; N/A, not applicable; URM, under‐represented minority.

Data for all variables were available for 1365 visits from 659 individual participants (average 2.1 visits/participant, range 1–7). Mixed‐effects logistic regression modeling revealed that BMI and CDR global score are independently associated with phlebotomy failure. Specifically, participants had $\approx$10% greater odds of having a failed phlebotomy for every kg/m² unit increase in BMI and $\approx$50% greater odds for every increase in CDR global score relative to 0.0 (normal cognition) up to more than 4‐fold higher odds in participants with CDR 2.0 (moderate dementia) and 3.0 (severe dementia). Additionally, URM participants had twice the odds of suffering a failed phlebotomy relative to non‐Hispanic White participants, although this association did not reach statistical significance due to insufficient statistical power (P = 0.059; see Table 2 and Table S1 in supporting information). Conversely, age, sex, and education were not significantly associated with phlebotomy failure (Table 2). Treating education as a continuous variable rendered very similar results (not shown).

TABLE 2.

Factors associated with phlebotomy failure in the Massachusetts Alzheimer's Disease Research Center Longitudinal Cohort Study.

	OR	SE	95% Cl	z	p value
Visit age (years)	1.01	0.01	0.99, 1.03	0.70	0.482
Female sex	1.09	0.30	0.63, 1.88	0.31	0.758
Education > 12 years	0.93	0.35	0.44, 1.95	−0.20	0.841
URM	2.07	0.80	0.97, 4.41	1.89	0.059
BMI	1.10	0.03	1.05, 1.16	3.83	<0.001
CDR global score:	1.52	0.23	1.13, 2.03	2.80	0.005
CDR 0.5	1.49	0.44	0.84, 2.65	1.36	0.175
CDR 1.0	2.00	0.86	0.87, 4.63	1.62	0.105
CDR 2.0	4.68	2.97	1.35, 16.30	2.43	0.015
CDR 3.0	4.46	4.29	0.68, 29.38	1.56	0.120

Notes: Results are from the full mixed‐effects logistic regression model. Statistically significant results are in bold font whereas statistical trends are italicized.

Abbreviations: BMI, body mass index; CDR, Clinical Dementia Rating scale; CI, confidence interval; OR, odds ratio; SE, standard error; URM, under‐represented minority.

4. DISCUSSION

We found that higher BMI and more severe cognitive impairment as assessed with the CDR global score were independently associated with increased odds of phlebotomy failure by CRCs. Additionally, there was a non‐significant trend toward URM participants having a 2‐fold increase in the odds of phlebotomy failure compared to non‐Hispanic White participants, whereas age, sex, and education did not independently impact the occurrence of phlebotomy failure.

In a meta‐analysis of risk factors for difficult peripheral intravenous cannulation or first attempt cannulation failure, age was insufficiently reported and education was not reported, whereas female sex was identified as a risk factor only in the hospital ward setting. ¹¹ In contrast, reduced visibility and palpability of veins, recorded as “difficult veins” in our study, are known major predictors of phlebotomy failure. ¹² , ¹³ , ¹⁴ Notably, two of the factors associated with higher odds of phlebotomy failure in this study are related to lower visibility and palpability of veins. Greater BMI can impair vein visibility and palpability due to thicker subcutaneous tissue and greater vein depth, whereas vein visibility can be reduced in URM versus non‐Hispanic White individuals due to darker skin. ¹¹ , ¹⁵ Thus, the factors associated with phlebotomy failure by MADRC CRCs closely resembles those reported for professional phlebotomists and nursing staff. The positive association between phlebotomy failure and increasing CDR global score could be explained by the neuropsychiatric manifestations (e.g., anxiety, agitation), frailty, and/or dehydration often associated with dementia progression. Of note, this association was independent of BMI, which is known to lower with disease progression.

Technological advances may help overcome these challenges in these subsets of participants. Near‐infrared vein illumination devices can improve both older adults’ venipuncture experience ¹⁶ , ¹⁷ and operator's subjective venipuncture skills. ¹⁸ Point‐of‐care real‐time ultrasound is commonly used in inpatient settings to reduce the total procedure time, number of needle insertion attempts, and number of needle redirections in patients with difficult peripheral venous access, ¹⁹ but its availability in outpatient research spaces is usually limited. Dried blood spot assays using fingerstick blood are another promising avenue under development. ²⁰ The tolerability of all these alternative strategies in moderate and severe dementia stages remains to be investigated.

Strengths of this study include the large sample of older adults spanning the full cognitive spectrum from normal cognition to dementia and the standardized data collection using the NACC UDS. Several limitations should also be noted. First, the cohort's high education attainment may have attenuated the association between low education and the odds of phlebotomy failure. Second, the limited number of participants from URM groups precluded meaningful comparisons between Hispanic/Latino and non‐Hispanic Black participants. Third, although MADRC CRCs typically remain in their positions for ≈ 2 years and likely improve their phlebotomy proficiency over time, we were unable to account for CRCs’ experience because the years of experience were not systematically collected. Finally, the generalizability of our findings to other ADRCs or AD/ADRD research cohorts may be limited by characteristics unique to the MADRC's cohort and research infrastructure.

In summary, greater BMI and cognitive impairment severity as well as URM status were associated with higher odds of phlebotomy failure in the MADRC Longitudinal Cohort Study of cognitive aging. Strategies to maximize participation of these subpopulations in blood‐based biomarker and genetics research are needed to improve representativeness.

CONFLICT OF INTEREST STATEMENT

The authors declare no conflicts of interest. Author disclosures are available in the Supporting Information.

CONSENT STATEMENT

All participants and/or their legal representative or appointed guardian provided informed written consent to participate in the MADRC Longitudinal Cohort Study under the MGH Institutional Review Board protocol #1999P003693 and a separate informed written consent to undergo phlebotomy and authorize the research use of their blood sample under the MGH Institutional Review Board protocol #2006P002104.

Supporting information

Supporting Information

Supporting Information