Stop pulling from the floor. Start pulling from 22 centimetres up.

Most lifters past 45 who quit the deadlift after a back tweak don't have a technique problem. They have a tool-fit problem.

The conventional floor pull stacks three simultaneous mobility demands on a desk-conditioned body. Ankle dorsiflexion. A deep hip hinge with a neutrally braced lumbar spine. Full bracing at the exact moment the extensor moment is maximal. Swinton and colleagues⁸ measured what changes when you swap the tool. In 19 trained powerlifters tested at submaximal loads from 10 to 80 percent of one-rep max, the hexagonal (trap) bar produced significantly higher peak forces, velocities and powers than the straight bar, with significantly lower peak lumbar extensor moments and significantly higher peak knee extensor moments. The mechanical signature shifts from a spine-loading, posterior-chain dominant pattern toward a more centred, shared knee and hip pattern. The data says what most coaches won't say out loud. For the 45-plus executive, the default deadlift is not the conventional floor pull. It is the trap bar, started around 22 centimetres off the floor, loaded progressively.

Two orthodoxies dominate the conversation and both read the evidence wrong. The gym-bro line says the deadlift is the king of exercises, age-agnostic, and that technique is the only variable worth tuning. The cautious-physio line says that past 45 the conventional deadlift grinds the lumbar spine and should be abandoned. The lumbar-biomechanics literature since Cholewicki, McGill, and Norman³ and Granhed, Jonson, and Hansson⁵ reads differently. The adult spine tolerates and adapts to heavy compound loading when the variant, starting height, and progression are aligned with the lifter's available mobility. The alignment is the entire question. Scope: this article addresses healthy 45-plus adults without diagnosed lumbar pathology, acute disc injury, or recent spinal surgery. Those situations require clinical supervision, not self-programming.

What the studies actually say

Three findings organize the deadlift biomechanics literature for the 45-plus reader. Hold these three, and most of what gets argued in coaching forums on deadlift safety, variant choice, and starting height reads as noise.

Finding 1. The trap bar reduces peak lumbar moment at matched or higher load. Swinton, Stewart, Agouris, Keogh, and Lloyd⁸ ran a direct biomechanical comparison of straight and hexagonal barbells in 19 trained powerlifters across loads from 10 to 80 percent of one-rep max. The hexagonal bar produced significantly higher peak forces, velocities and powers, and significantly lower peak moments at the lumbar spine and hip, while peak knee extensor moments were significantly higher. The mechanical picture is consistent. Load gets shared across the hip, knee, and spine instead of concentrating at the lumbar extensors. Camara, Coburn, Dunnick, Brown, Galpin, and Costa² measured muscle activation and power with both bars in 20 resistance-trained men across 65 and 85 percent of one-rep max, and reported significantly greater peak power and peak force with the hexagonal bar, significantly greater vastus lateralis activation, and significantly lower erector spinae activation on the eccentric phase for the hexagonal bar (normalized EMG around 0.61 versus 0.75 for the straight bar). The direction converges with Swinton. Cholewicki, McGill, and Norman³, in the canonical paper on lumbar spine loads during extreme weightlifting, modeled lumbar spine forces during heavy deadlifts in competitive powerlifters and reported peak L4-L5 compression values on the order of 17 000 N in the heaviest lifts. The paper frames these values as exceeding thresholds associated with acute spinal injury, and notes that the lifters sustained them only after years of progressive training. The practical reading is not that the adult spine is routinely operating at 17 000 N, it is that the tolerable ceiling is not a function of age but of cumulative progressive loading.

Finding 2. Starting height is a legitimate lever, not a compromise. Escamilla, Francisco, Fleisig, Barrentine, Welch, Kayes, Garhammer, and Andrews⁴ ran a three-dimensional biomechanical analysis of sumo versus conventional deadlifts in 24 lifters (12 sumo and 12 conventional) at 70 percent of one-rep max. Sumo lifters used significantly wider stances with greater foot turnout, and the conventional style produced greater vertical bar displacement and mechanical work, while hip moments did not differ significantly between the two styles. The principle that carries beyond this specific comparison is mechanical. The position of the bar at the start of the lift changes the ankle, knee, and trunk angles at the break, which in turn modifies how much extension demand the lumbar spine must tolerate at that phase. Raising the start height from the floor to mid-shin or knee-level reduces the hip flexion required at the break, and by extension reduces the lumbar extensor moment at that phase of the lift. The exact magnitude at each increment has not been quantified in a single peer-reviewed trial, so the biomechanical reasoning is presented as a directional default, not as a single-study claim. McGill⁷, in his core-training evidence review, anchors the principle by describing the hip hinge as a hip-dominant movement with a neutral spine, and emphasizes that peak shear and compression on the lumbar spine occur at the positions where the spine is most flexed under load. Raising the start height protects the neutral spine at the point of highest shear demand.

Finding 3. The deadlift remains trainable even with a low-back history, under the right conditions. Berglund, Aasa, Hellqvist, Michaelson, and Aasa¹ ran 8 weeks of deadlift training in 35 participants with mechanical low back pain of at least three months' duration and reported significant improvements in pain intensity, activity limitations, and function. A subgroup analysis identified lower baseline disability, lower baseline pain intensity, and higher trunk-extensor endurance (Biering-Sørensen test) as the clearest predictors of response. Welch, Moran, Antony, Richter, Marshall, Coyle, Falvey, and Franklyn-Miller¹⁰ ran a single-arm free-weight resistance intervention including deadlift in 30 adults with chronic low back pain and reported significant reductions in pain and disability, alongside MRI-defined reductions in paraspinal fat infiltration at L3-L4 and L4-L5. The sample is modest, the design is single-arm, and the direction is encouraging rather than proof of effect. Granhed, Jonson, and Hansson⁵ measured lumbar bone mineral content in 8 competitive powerlifters and reported values well above those of matched controls, with a strong correlation (r-squared around 0.82) between bone mineral content and training loads. The spine adapts to chronic heavy loading. It does not degrade under it.

See the full evidence base for every study referenced here.

Is deadlifting safe for your lower back after 45?

Yes, deadlifting is safe for your lower back after 45 in healthy adults without diagnosed lumbar pathology, when the variant, starting height, and progression match available mobility and lifting history. The evidence from Berglund 2015¹ and Welch 2015¹⁰ shows that programmed deadlift training is not only safe but also reduces pain and improves function in specific subgroups of chronic low back pain when introduced progressively. Granhed 1987⁵ shows that the spine adapts to heavy chronic loading with higher bone density, not degeneration. The real question is not deadlift or no deadlift. It is which variant, at what starting height, with what progression. Scope note: diagnosed lumbar pathology, acute disc injury, radicular symptoms (pain, weakness, or sensory changes radiating into the leg), or recent spinal surgery all sit outside self-programming. Those cases require clinical supervision, ideally from a physiotherapist with strength training competence.

Technique is a variable. It is not the dominant variable.

Technique isn't the lever here. Variant and starting height are. The deadlift technique literature reads technique as the lever worth tuning; for a 45-plus desk-conditioned executive, that reading is incomplete. The conventional floor pull requires three separate mobility components to converge simultaneously at the break of the floor: ankle dorsiflexion above a certain threshold, deep hip flexion with a neutrally braced lumbar spine, and a full thoracic brace at the highest extensor moment of the lift. Eight to twelve hours of daily sitting shortens the posterior chain and restricts dorsiflexion. The hip hinge pattern becomes an unpractised movement. Under submaximal load on a Tuesday at 7 AM, the body looks for the compliance it has, and it finds that compliance in a lumbar spine that flexes rather than in hamstrings that lengthen. The technique cue arrives too late, because the body is already past the position where the cue could have worked.

The practical reading is not that technique doesn't matter. It matters. It is that in a given session the most load-bearing variables are the variant selected and the starting height chosen. Technique layered on top of the wrong variant or the wrong height is technique pointed at an unsolvable problem. The squat article treated the analogue problem for the squat pattern, where variant choice dominated depth as the driver of joint-friendly loading. The parallel for the hinge pattern is the same. Variant and starting height carry the first eighty percent. Technique refinement carries the last twenty. Getting the order right is the point.

The three mechanisms that actually matter

Three mechanisms explain where the 45-plus deadlift tends to degrade. One dominates in practice, one is the amplifier that gets underweighted, and one is the misattribution that masks what is really happening.

Mechanism 1. The floor-break stacking, the biomechanical driver. The conventional deadlift at the floor requires, in the same instant of the break, ankle dorsiflexion past a practical threshold, a fully flexed hip with the lumbar spine held neutral under load, and a thoracic brace at the peak demand of the lift. Cholewicki 1991³ documents that competitive powerlifters at true heavy loads reach lumbar spine compression values on the order of 17 000 N, values the paper frames as exceeding thresholds associated with acute spinal injury and sustained only after years of progressive training. Granhed 1987⁵ documents the long-term bone mineral signature of that adaptation, with values well above matched controls. The spine that has not held the floor-break position under any meaningful load in fifteen years is operating at the edge of its current tolerance, not its theoretical ceiling. The direct lever is to remove the requirement for full dorsiflexion and deepest hip flexion from the critical path. Either raise the starting height, or switch to the trap bar where the load is centred on the lifter rather than ahead of the shins.

Mechanism 2. Desk-conditioning, the amplifier. Eight to twelve hours of daily sitting shortens the hip flexors, restricts ankle dorsiflexion, and locks the thoracic spine in flexion. The hip hinge pattern is functionally unpractised for most executives, whose daily bending is lumbar flexion to reach for a coffee cup, not a hip hinge. The compliance the body finds at the bottom of an unloaded floor reach is lumbar rounding, not hamstring lengthening. The fix lives outside the training session as much as inside it. The mobility article details the ankle and hip work. Inside the session, the hip hinge pattern is practised unloaded before load is added: pelvis drives back, chest drives forward, weight stays on mid-foot, hands track the front of the thighs down to the knee.

Mechanism 3. Tool-fit error, the misattribution. The cultural reflex that the conventional deadlift is the only real deadlift still drives most default programming. For a 45-plus executive with compressed recovery, desk-conditioned mobility, and a fragmentary barbell history, the conventional floor pull is rarely the correct default. The statement that it is dangerous is too strong. The statement that it is the wrong default for this profile is exactly right. The trap bar and elevated pulls align the load path with the available mobility. Switching variant or starting height is not cheating. It is tool-fit done right. The executive who blames age or blames his knees or blames the exercise is almost always running the wrong tool for his current context.

The framework: double pivot, decision tree, 8-week progression

The double pivot

Two levers dominate. Pull them together.

• Variant. Trap bar over conventional, for the 45-plus executive default.
• Starting height. Roughly 22 centimetres over zero, for the 45-plus executive default.

The rule that combines both: start with the trap bar at around 22 centimetres. Lower the starting height toward 15 centimetres, then toward the floor, only after four or more weeks of clean three-point checks at each preceding height.

The 3-point technique check

Thirty seconds, before each hinge session.

1. Hip hinge test. Feet parallel, hands travel down the front of the thighs. If the lumbar spine visibly rounds before the hands reach the knees, the hinge pattern, not force, is the limiter. Practise unloaded before adding load.
2. Neutral spine under breath. Set up in the pull position, hold static for 10 seconds at the starting position, film from the side. Any visible lumbar flexion means reduce load or raise the starting height.
3. Brace test. Diaphragmatic inhale and 360-degree brace, 5-second hold under the working load. If the brace collapses at the break from the floor, the demand exceeds the brace currently practised. Raise the starting height.

Two failures out of three, and the session downshifts. The bar is intentionally low.

The 5-variant decision tree

Indexed on joint history, equipment, and training phase.

The sequencing rule: most 45-plus executives anchor their hinge week around the trap bar plus the Romanian deadlift, with the conventional reserved as a peaking tool one training cycle in three rather than a weekly default. The sumo enters as a specialty when hip abduction and external rotation are available and the lumbar lever is the dominant concern. The goblet is the entry tool and the travel tool; grip becomes the limiter before the hips do above roughly 30 kilograms.

The 8-week progression

Two sessions per week, integrated with the 3-day rotation and staying within the 10-to-15 weekly set ceiling for the posterior chain.

• Weeks 1-2. Trap bar from 22 centimetres, 3 sets of 8 at RPE 6 to 7. Goal is the hinge pattern and lumbar tolerance, not the load.
• Weeks 3-4. Trap bar at 22 centimetres, 4 sets of 6 at RPE 7. Add the Romanian deadlift, 3 sets of 10 at RPE 7, for hinge specificity without peak compression.
• Weeks 5-6. Trap bar drops toward 15 centimetres only if all three checks are clean. 5 sets of 5 at RPE 7 to 8, 2 to 2.5 kilograms micro-progression per week.
• Weeks 7-8. Trap bar at 15 centimetres, or from the floor if mobility and checks permit, 5 sets of 3 at RPE 8 plus a back-off of 2 sets of 8 at RPE 7. Deload in week 9.

Rack pulls above the knee are overload, not deadlifts

Rack pulls from around 40 centimetres (above the knee) are useful for grip and lockout overload at supramaximal loads, but they do not program the full hinge. Log them as an accessory. The main hinge personal records live on 0 to 30 centimetre variants.

Returning from a tweak, grip, and bracing

Three adjacent variables deserve a paragraph without opening a separate book.

Return from a back tweak. Berglund 2015¹ and Welch 2015¹⁰ support deadlift-based return protocols for chronic low back pain when baseline pain intensity is low and trunk-extensor endurance (Biering-Sørensen) is adequate. A practical return ramp goes goblet, Romanian deadlift unloaded, Romanian deadlift lightly loaded, trap bar from 22 centimetres, then back into the 8-week progression. Allow four to six weeks after the tweak resolves before reintroducing any floor-level pull. Persisting pain past 10 days, or any neurological signs (radiation into the leg, weakness, or sensory changes), means a clinical consult before any lifting. A physiotherapist with strength training competence is more useful than a generalist coach for this transition.

Grip as the hidden limiter. Past 45, grip strength becomes a sub-system worth auditing before concluding that the hips or hamstrings are the bottleneck. Straps for occasional sets (not every set) remove the grip variable. Farmer's walks twice a week develop the grip independently and also track a marker associated with healthspan.

Bracing and the diaphragm. A non-maximal 30 to 50 percent brace is adequate for the majority of sets under RPE 7. Full Valsalva is reserved for the top sets at RPE 8 to 9. Lifters with diagnosed hypertension should clear regular Valsalva with their physician before programming it as a default.

What to measure, what to ignore

Three numbers. Everything else is tracking theatre.

1. Main hinge variant load. Trap-bar 5-rep max or 8-rep max, logged session by session. The 8-week trend is the signal. Day-to-day variation is noise.
2. Morning low-back 0 to 3, during the 48 hours after a heavy hinge session. Two consecutive days at 2 or above means deload the hinge for that week, not push through.
3. Monthly hip hinge wall test. Feet 20 centimetres from a wall, hinge until the glutes touch the wall. Measure the vertical distance from hands to floor at the touch point. The number shortens as hip flexibility and hinge pattern consolidate.

What to ignore: one-rep max deadlift tested more than once every 16 weeks (neural noise dominates the signal), bar path capture (noise above signal for non-competitive lifters), subtle hip-shift observation (observer variance dominates), and rack-pull-above-the-knee personal records (overload, not a full hinge).

The heavy-loading premise behind this protocol is grounded in the LIFTMOR⁹ and LIFTMOR-M⁶ trials. Watson 2018 delivered high-intensity resistance and impact training to postmenopausal women with osteopenia and osteoporosis for 8 months and reported significant gains in lumbar spine and femoral neck bone mineral density and functional measures, with only one minor adverse event (a lower-back spasm) reported in the HiRIT intervention arm over eight months. Harding 2020 replicated the direction in middle-aged and older men with low bone mass. Heavy compound loading, sequenced well, remains productive and safe well before the traditional cautious-physio cutoff.

The bottom line

The deadlift isn't dangerous past 45. The conventional floor pull as a weekly default often is. The hinge pattern remains a pillar of longevity, posterior-chain strength, grip capacity, and bone density. What changes is the tool. The double pivot is trap bar and 22 centimetres. The rest (Romanian deadlift, sumo, conventional) rotates in for specificity and peaking. Pull from 22 centimetres up.